v2.0.0

Fixed

• Reported cell count off by -1 in report summary table.
• Issue with TSV concat/splitting during combine_bam_and_tags stage.
• Issue introduced in v1.1.0 that caused a partial BAM file to be output.
• Corrected example command in README.
• Incorrect reporting of unique gene and transcripts in report table.
• Processed expression matrix entries incorrectly filtered.
• Gene identity of multimapping reads could be incorrectly assigned.

Changed

• Read chunking done in library code.
• --process_chunk_size parameter changed to --fastq_chunk
• Resource declarations in Nextflow processes.
• Simplified read batching and decoupled from CPU usage parameters.
• Expression matrix construction code reworked to reduce memory usage.
• Adapter search step now 3x faster.
• Barcode assignment 3x faster.
• Feature assignment now 15x faster.
• UMI clustering 20x faster.
• UMAP creation memory use reduced 6-fold and up-to 30x faster (and
  always enabled).
• Final read tagging step is 3x faster.
• Combined various preprocessing steps into a single process to avoid
  unnecessary file writes.
• Updated stringtie2 to v2.2.2.
• Pre-calculate report summary data to reduce disk-space and IO overheads.
• Single BAM per-sample is now always produced (option --merge_bam is removed).

Removed

• Several workflow parameters as part of resource management simplification.
• --plot_umaps option, as UMAP generation has been made much more efficient
  and is always enabled.
• --merge_bam option.