H5pyd

Feature: hsload now will import any custom filters defined on a HDF5 dataset to the filter pipeline for the HSDS dataset.
If the filter is supported in HSDS (e.g. bitshuffle, blosclz, lz4, lz4hc, zlib, zstd as of HSDS v 0.8.5), the filter will be invoked on reads and writes to the dataset, otherwise it will be ignored.

Updates:

* Fixes for numpy deprecation issues
* Support for creating domains in root folder (for admin users)
* Removed h5serv support from test cases
* Fix for issue: 152

Fixes:
* apply environment variables (HS_USERNAME, HS_PASSWORD, HS_BUCKET) when present
* Fix hsload -a --noclobber behavior

Features:
* display filters in hsls (144)
* add --ignore-filters option to hsload (ignore any existing filters in source hdf5 file) (143)

Fixes
* enable hsload for non-utf8 encodable attributes (135)
* enable hsload with compact datasets (136)
* enable hsload with datasets using object reference type (139)
* emit warning when hsload is used with datasets with the scale offset filter (142)
* use enviornment variable override for hs tools (145)

Features:

* cmdline tools use unified help system for more consistent display for help screens
* hsstat tool added (same functionality as old hsinfo <filename>)

Added aggregation options (-a, --extend, --extend-offset) for hsload. Run hsload -h for usage (see 120)
dataset.chunks will always return None or tuple (see 130)
fix for FLETCHER32 filter (see 128)
fix for compound types with one field (see 127)
fix for hsload on empty dataset (see 116)
hsload will exit on first failure unless --ignore flag is used (see 111)