Updates for May 9

[bfhealy]

Pull requests

• Add nested ELS_ECE_EAOV period algorithm with minimal extra compute #362
• Revise docs, delete another .DS_Store #363
• Add 0.2d to period-finding mask #364
• Update slurm script generation with new arguments #365
• Pin periodfind 0.0.5 #366
• Fix failing test suite by removing melman references #368
• Add additional hyphen to arguments in download/upload/annot scripts #369
• Add ability to download active learning classifications, merge with training set #370
• Update usage.md to be clearer about named arguments #371

Issues

• Incorporate XGBoost algorithm into SCoPe #194
• Inconsistency between Lomb-Scargle CPU and GPU algorithms #360
• Re-implement LS/CE/AOV period-finding sequence #353
• Modify period-finding approach to obtain more reliable periods #341
• EAOV_periodogram algorithm should use max statistic to identify best period #358
• Fix failing test by removing melman references #367
• Code/strategy to add active learning classifications to training set #211

Going forward

We will compare results of running the GPU period algorithms on DR5 sources vs. their existing periods. We will also work on:

• Expanding the capabilities of the code that checks Kowalski collection completeness
• Adding a hyperparameter grid search to DNN training