Biothings-client

Fix a bug when certain query parameters are passed in as lists. (12)

Use doc_type key in metadata to generate client without having to specific biothings_type (6).
Have a more consistent user-agent pattern (7).
Fixed the issue when passed query terms contains white spaces in querymany method (8).
Added warning when as_dataframe set to True and fetch_all is True (9).

Added client-specific doc-strings to match with current mygene.py and myvariant.py clients.
Enhanced support for MyChem.info client (using `get_client("chem")` or `get_client("drug")`)
Added initial support for MyDisease.info client.
Dropped the support for Python 2.6 (might still work but we don't explicitly test it under
Python 2.6 any more)
Support for Python 3.7
When returned objects as a Pandas Dataframe (i.e. using "as_dataframe=True"),
duplicated and missing hits are now returned as dataframes as well (Thanks sdhutchins!)
Many code optimizations.

Optimized the code folder structure

initial release



RELEASE

This is the procedure we use for "biothings_client" package release

1. requires both `build` and `twine` packages installed

bash
pip install build twine


2. Update version number in [base.py](biothings_client/base.py).

3. Check and update [pyproject.toml](pyproject.toml) if needed (dependencies, metadata etc.).

4. Build the package locally:

bash
python -m build


5. Test the package built locally:

bash
pip install dist/biothings_client-0.3.0-py2.py3-none-any.whl


And run any local test as needed.

6. Prepare github repo for the release:

* Create a tag for each released version (with "v" prefix):

bash
git tag -a "v0.3.0" -m "tagging v0.3.0 for release"


* If everything looks good, push to the remote:

bash
git push --tags


7. Upload to PyPI:

bash
twine upload dist/*


Note: this step needs to be done by newgene under ["newgene" PyPI account](https://pypi.org/user/newgene/) or any authorized PyPI user.