Download  no longer tries to get existing files

See merge request !329

Scores loading fixes

Closes #194 and #196

See merge request !328

When loading scores, the 'probe_template_id' column is now the correct
way to differentiate between probe templates. Caused issues with
pipelines not even outputting the 'probe_key' column anymore.

When loading a score file, indicate to pandas to interpret the score
column as float (prevent issues when the first elements are empty).

meta (conda): fix the entry-points not being created correctly with conda

See merge request !324

docs: remove refs to bob.extension and clean deps.

See merge request !323

feat: retry a download if checksum is not correct

See merge request !327

Adds `checksum_mismatch_download_attempts` parameter to `download_file`.
Adds a loop when downloading that checks the new file checksum.

Fixes #195.

feat: add an option to select the score column in the bob bio measure commands

See merge request !325

fix: stop T-norm fit from exploding in memory.

See merge request !326

Removed the unnecessary copy of `data` and storage into a Sample object.
Now saves the stats in a dict that is picklable.

Added support for specifying the CSV column containing the scores.

meta(deps): add bob as dependency in new structure

See merge request !322