v0.3.3

What's Changed

• Assorted small doc fixes by @delucchi-cmu in #310
• User arguments to retain intermediate files by @delucchi-cmu in #311
• Apply read_columns kwarg in parquet reader by @troyraen in #315
• Get all file_reader provenance info programmatically by @troyraen in #316
• Spruce up the way we check for original input files. by @delucchi-cmu in #314
• Fail fast on reduce stage failures. by @delucchi-cmu in #317
• Print file name for mapping/splitting failures. by @delucchi-cmu in #318
• Explicitly depend on scipy by @delucchi-cmu in #320
• Redact storage options, as this may include user keys. by @delucchi-cmu in #322
• Resume functionality for margin cache pipeline. by @delucchi-cmu in #321
• Tidy up some resume functionality by @delucchi-cmu in #324
• Use the original catalog schema in more reads/writes. by @delucchi-cmu in #325
• Use a subset of pixels when creating a margin cache. by @delucchi-cmu in #323

Full Changelog: v0.3.2...v0.3.3