Marrow.mongo

General corrections and changes:

* Array field could not be filtered as not-equal. 43
* The new `Lockable` trait implements mutex lock behaviour at the document level. 47
* Package build and automated testing adjustments including expanded build matrixes, Bandit exclusions, and multiple MongoDB versions.
* Dead code removal.
* Adaption to future reserved word use; `await` will be reserved in Python 3.7+.
* Corrections for certain edge cases involving casting of `None` values.

Field types now available:

* `Link` for the storage of URI values such as HTTP URLs, `mailto:`, `tel:`, etc. 45 / 39

* `Mapping` field to automatically perform read-only translation of a keyed list of embedded documents into a dictionary. 46

* `Set` field will utilize a true `set` instance Python-side.

Field enhancements:

* Fields may now be excluded from positional instantiation. 28

* Fields may now be adapted / mutated to specialize when inheriting without complete replacement. 38

* `Alias` fields may now trigger deprecation warnings if requested. 48

* `Date` fields are now timezone aware if `pytz` is installed, and able to intelligently utilize the server-local timezone if `tzlocal` is installed. (Or just utilize Marrow Mongo's `tz` installation flag.) 51

* `PluginReference` can now perform simple search and replace in Python import references, allowing for mapping of old import paths to new ones during code refactoring. 49

__Please note that due to Pypi stupidity, version `1.1.1.1` there is actually `1.1.1`.__

New or updated in this release:

* Removal of diagnostic information and updated testing/commit configurations, improving commit performance and bumping Pypy3 versions.
* Correction of ABC participation (and missing shallow copy method) for Pypy use of query fragments. 32
* Corrected `$regex` generation.
* Collation support.
* Passing an existing document (with `_id` key) to an `ObjectId` field will utilize the ID provided therein. 20
* Enhanced `String` field capabilities to include stripping and case conversion. 33
* Shared `utcnow` helper function.
* Improved documentation coverage.
* Improved generalized programers' representations.
* Improved query fragment merging.
* Corrected Reference field behaviours.
* Dead code removal.
* Updated `Array` and `Embed` field default value handling to reduce boilerplate.

**Potentially backwards-incompatible changes:**

* Simplification to only support a single referenced kind in complex fields such as `Array` and `Embed`. As multi-kind support was not fully implemented, this should not disrupt much.

New fields, including:

* `Decimal` — 23
* `Period` — Storage of dates rounded (floor) to their nearest period.
* `Markdown` — Rich storage of Markdown textual content. 34
* `Path` — Store a PurePosixPath as a string. 35

Traits are new, see 26, including:

* `Collection` — Isolating collection management semantics from the core `Document` class.
* `Derived` — Isolating subclass management and loading from the core `Document` class.
* `Expires` — Automated inclusion of `TTL` (time-to-live) field and index definitions, including expiry check on load.
* `Identified` — Isolation of primary key management from core `Document` class.
* `Localized` — Management of contained localizable top-level document content.
* `Published` — Management of publication/retraction and dedicated creation/modification times.
* `Queryable` — Encapsulation of collection-level record management. (**Not** an Active Record pattern.)

New in this release:
- Add Landscape.io integration.
- Improve overall code health. (14)
- Added missing project metadata.
- Updated installation documentation. 81e7702
- Remove dependency on `pytz`. 815a74a
- Removed our own `compat` module; schema already has a sufficient one.
- Allow for `Reference` fields to cache data they reference. (8)
- `Array` & `Embed` dereferencing + `Alias` pseudo-field support. (12)
- Ability to dereference `Array` and `Embed` subfield values when querying through class attribute access.
- Added `Alias` pseudo-field to allow the creation of shortcuts for value retrieval and assignment (via instance attribute access) and querying (through class attribute access).
- `Array` and `Embed` now persist their typecasting within `__data__`, to preserve changes to nested values. (This is generally safe, however do not utilize `PluginReference` as an embeddable kind.)
- Allow for fields to be combined, not just query documents. (11)
- Field references (`Q` instances generated through class-based attribute access of fields) may now be combined to save time in queries involving multiple fields being compared against the same value.
- Parameterized filter, sort, projection and updates. (4)
- Addition of `~` inversion / `$not` support on `Ops`.
- Split `Ops` types.
- Ensure Document uses `odict`.
- GeoJSON and geographic querying support. (6)
- Added Document types:
- `GeoJSON`
- `GeoJSONCoord`
- `Point`
- `LineString`
- `Polygon`
- `MultiPoint`
- `MultiLineString`
- `MultiPolygon`
- `GeometryCollection`
- Added field query operators:
- `near`
- `intersects`
- `within`
- Added parametric filter operators:
- `near`
- `within`
- `within_box`
- `within_polygon`
- `within_center`
- `within_sphere`
- `intersects`
- Ability to perform certain collection-level operations. (17)
- Added Document class methods:
- `create_collection`
- `get_collection`
- `create_indexes`
- Added the following Document class attributes to control collection settings:
- `__collection__` - the name of the collection to use
- `__read_preference__` - default ReadPreference
- `__read_concern__` - default ReadConcern
- `__write_concern__` - default WriteConcern
- `__capped__` - the size, in bytes, to allocate as a capped collection
- `__capped_count__` - additionally limit the number of records
- `__engine__` - override storage engine options
- `__validate__` - one of 'off' (the default), 'strict', or 'moderate'.

- Initial release of basic field mapping functionality.