ENH: L1L2 option, fixed alr idx and relative abundances

[adamovanja]

this PR:

• adds L1 and L2 regularisation options for linear regression through ElasticNet model from ScikitLearn
• adds fixed alr index to use for all trials to most non-zero feature in train_val
• enforces feature table to use for modelling to be of relative abundances (not absolute counts)
• adds model type as metric to MLflow

[codecov]

Codecov Report

Attention: Patch coverage is 99.45055% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 62.71%. Comparing base (54a62b6) to head (822d222).

Files	Patch %	Lines
q2_ritme/tune_models.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #14      +/-   ##
==========================================
+ Coverage   57.60%   62.71%   +5.10%     
==========================================
  Files          17       17              
  Lines         920     1046     +126     
  Branches       92       94       +2     
==========================================
+ Hits          530      656     +126     
  Misses        385      385              
  Partials        5        5              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.