Notebook unittests for Tellseq B and Tellseq C notebooks

[AmandaBirmingham]

• switch tellseq B notebook to using string constant MINIPICO_LIB_CONC_KEY instead of string literal to reference dataframe column name
• modify tellseq B notebook to support unit tests
• add unit test for tellseq B notebook main path
• update tellseq B test files to reflect change in add_controls output
• modify tellseq C notebook to support unit tests
• add unit test for tellseq C notebook main path
• update tellseq C test files to reflect change in add_controls output

This PR was an absolute beast because of differences in last-digit floating point representations in the output files created on my development machine versus those created in the github CI. To reign this in, I also made the following changes:

• modify number of digits for INPUT_DNA_KEY column in dataframe output file of tellseq A notebook
• modify number of digits for MINIPICO_LIB_CONC_KEY column in dataframe output file of tellseq B notebook
• update tellseq A, tellseq B, and tellseq C dataframe test files to reflect digit limits

[coveralls]

Pull Request Test Coverage Report for Build 20936646009

Details

• 18 of 20 (90.0%) changed or added relevant lines in 2 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.1%) to 92.743%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
notebooks/tests/test_tellseq_B_concentration_estimation.py	9	10	90.0%
notebooks/tests/test_tellseq_C_equal_volume_pooling.py	9	10	90.0%

Totals
Change from base Build 20929950507:	0.1%
Covered Lines:	6607
Relevant Lines:	7124

---

💛 - Coveralls

[antgonza]

Just one minor question.

[antgonza]

Out of curiosity, why 10?

[AmandaBirmingham]

An excellent question! I have two answers:

1. I asked chatgpt how to handle having floating point numbers expressed weirdly in file output (like 0.1 being written out as 0.09999999999999) and it suggested formatting the dataframe write out, saying "%.10g gives up to 10 significant digits and avoids lots of trailing noise; it’s a good general default for “human-ish” TSV."
2. I implemented that and then looked at what changed in the output file, and what I saw was that the changes were cutting off digits I did not believe in the first place, like 7.499350000000001 becoming 7.49935, 7.504912999999999 becoming 7.504913, 7.4984139999999995 becoming 7.498414, and so forth.

That said, if you feel that it should be other, I am definitely not married to this. It does seem to matter what the particular data column is (based, I assume on how it is calculated) whether it has a bunch of meaningful digits or not.

[antgonza]

Thank you for the explanation, sounds good to me.