Pdr

Added
Features
- FITS files can now be opened 'directly' (without a PDS label).
See [README.md](https://github.com/MillionConcepts/pdr/blob/main/README.md) for more detailed usage.
- 1d, non-structured ARRAY objects are now compatible with `dump_browse`
- support for ARRAY objects with nested ARRAY or COLLECTION objects
- [`pytest`-compatible unit test suite](https://github.com/MillionConcepts/pdr/blob/main/pdr/tests/)
intended to complement our [comprehensive regression test framework](https://github.com/MillionConcepts/pdr-tests)

Dataset Support
- most New Horizons datasets
- Venera 15 and 16 derived data
- Giotto PIA and VEGA 1/2 Puma_mode data
- Venus Climate Orbiter data
- Saturn Ring Plane Crossings (RPX) 1995-1996 data
- most Phoenix datasets
- Deep Impact and EPOXI data
- Mars Pathfinder data
- several MRO datasets including HiRISE EDRs, MCS, MARCI, and CTX
- see [supported_datasets.md](https://github.com/MillionConcepts/pdr/blob/main/supported_datasets.md) for details.

Changed
- Debug-mode error tracking is substantially more detailed
- Debug-mode error logs are now written as JSON
- Assorted backend refactoring, cleanup, error checks, and in-code documentation

Fixed
- Fixed incorrect PDS3 object -> FITS HDU mapping for several data types
- scaling/offset defined in FITS tables (BSCALE/BZERO) is now applied correctly to output `DataFrames`
- 8-byte integers found in some nonstandard binary files are now handled correctly
- more consistent handling of long / malformatted PVL comments
- improved handling of 'stub' primary FITS HDUs

Added
Features
- Support for FITS tables
- Documentation on behavior of FITS files (in README)
- Support for VAX floating-point numbers
Dataset Support
- Rosetta ALICE EDR, RDR, and REFDR data
- several Galileo datasets
- sampleninterleaved (BIP) images, including:
- DAWN VIR EDR and RDR cube products
- Rosetta VIRTIS EDRs
- Deep Space 1, NEAR, Stardust, and Stardust-NExT datasets
- various image products stored as VAX floats, including Pioneer Venus radar and IMIDR images
- IBM_REAL and EBCIDC data types, including Pioneer Venus SEDR
- LRO CRaTER EDR secondary science and housekeeping tables
- several Mars Odyssey datasets
- several Vega datasets

Changed
- TABLE and SERIES objects now apply label provided offset and scaling factors.
This has not yet been implemented for ARRAY objects or BIT COLUMNS.
- License file updated to incorporate license from the vendored vax.py module.

Fixed
- Tables with nested containers now read correctly; closes [issue 50](https://github.com/MillionConcepts/pdr/issues/50)
- ^STRUCTURE pointers inside a COLUMN/FIELD now correctly load the relevant format file

Removed
- Spaces in BIT_COLUMN DATA_TYPE values no longer raise `UserWarning`s

Fixed
- re-enabled the ability to import `pdr.Data` by adding it back to `__init__.py`

This release represents a major refactoring effort to reduce technical debt and decrease workflow complexity.
**The user-facing `pdr.read()` interface has not changed.**
Added
FEATURES
- `pdr.Data` initialized with `debug=True`
- Errors can be inspected at runtime by accessing `Data.loaders["OBJECT_NAME"].errors`.
- Tracking logs are also saved to the `pdr/.tracker_logs` folder.
directory of the installation folder. They show which functions `Data` objects
used during loading processes -- and if those functions were successful.
- handling for a wider array of ISIS-style "qube" data and metadata,
including side/back/bottom/topplanes (as long as they are along only one
image axis)
Dataset Support
- most THEMIS qube products
- TODO for cassini XDR image scaling functionality
- additional LRO datasets: CRaTER, LAMP, LEND, LOLA, Mini-RF, and Radio Science
- several Venusian datasets including: Magellan GVDR, Pioneer Venus,
"Pre-Magellan" products at the GEO node, and Earth-based radar observations.
- several Lunar datasets including: GRAIL, Lunar Prospector, MSX,
Kaguya/Selene, and Earth-based radar and spectroscopy data.- (most) THEMIS qube products
- TODO for cassini XDR image scaling functionality
- additional LRO datasets: CRaTER, LAMP, LEND, LOLA, Mini-RF, and Radio Science
- several Venusian datasets including: Magellan GVDR, Pioneer Venus,
"Pre-Magellan" products at the GEO node, and Earth-based radar observations.
- several Lunar datasets including: GRAIL, Lunar Prospector, MSX,
Kaguya/Selene, and Earth-based radar and spectroscopy data.

Changed
- reworked fundamental data loading workflow. `Data` class no longer contains
all the loader functions, they've been refactored in the `loaders` module.
- `formats.core.py` now only contains special case checking functions and is renamed to `formats.checkers.py`.
Other functions that were previously in it (like `pointer_to_loader` and `generic_image_properties`) have
moved to the `loaders` module
- changes to various special cases based on the data loading workflow refactor
- reworked image-loading flow for better handling of various band storage
types and pre/suffixes
- BIL images now retain original byteorder
- reworked cassini xdr special case for compatibility

Fixed
- discovered some rosetta VIRTIS and cassini UVIS product types were not
actually reading correctly, marked them out of support

Removed
- m3 special case module (deprecated by new image-loading flow)
- messenger special case module (deprecated by new data loading workflow)
- rasterio loading options for image data
- `check_special_case` has been removed and the special cases have been moved to functions
that more specifically targeted to the issues of the particular dataset rather
than overriding the entire workflow.

Added
- support for a variety of Magellan data (see [supported_datasets.md](https://github.com/MillionConcepts/pdr/blob/main/supported_datasets.md)
for specifics)

Changed
- sample types are now more permissive and allow spaces
- line endings are less strict and allow between 0-2 carriage returns

Fixed
- HiRISE EDRs are now only listed as notionally supported (rather than known
unsupported and notionally supported)

Removed
- special case for mgs-rss data with IEEE REAL sample types (now supported in core
functionality)

Added
Features
- This change log file
- Support for bit columns with ITEMS
- Browse images can now be output in grayscale
- handling for COMPRESSED_FILE objects
Dataset Support
- a variety of LRO Diviner, Cassini, and Huygens data products
(see [supported_datasets.md](https://github.com/MillionConcepts/pdr/blob/main/supported_datasets.md)
for specifics)

Changed
- pdr will now accept non-UTF-8 characters in detached PVL label / format files
- refactored special case checking for readability/maintainability
- assorted linting and in-code documentation edits
- labels will by default read in up to 1 MB of the file (previously 500 bytes). If
you're trying to get faster performance for shorter attached labels en masse, pass
the new --pvl_limit parameter. If you're not running large numbers of attached
label files in sequence this is largely irrelevant.

Fixed
- label comments that are left unclosed by the data providers no longer prevent
reading in the data
- group offset computations by offset not start byte (allows opening of Juno Jane V04
products, closes [issue 43](https://github.com/MillionConcepts/pdr/issues/43))
- bit columns are now split based on both start_bit and number of bits rather than
simply start_bit (this fixed a number of previously incorrectly read files,
and mirrors handling for other column types)
- `dump_browse` now properly accepts the `float_dtype` argument; reducing
bit depth can significantly reduce memory use when browsifying large arrays
- malformed labels with extra end-block statements no longer crash `BlockParser`
- JUNO JIRAM RDR special case now covers tables as well as images
- HEADER objects in compressed files now load properly
- DATA_SET_MAP_PROJECTION_CATALOGS no longer go to read_header (which does
not work on them)

Removed
- special-case handling for COMPRESSED_FILEs
- support for MSL CCAM LIBS EDRs