Release notes v0.16.25

Draft release notes for v0.16.25

These notes summarize the kfdevel line prepared for merge into master.

AMALGKIT is now Python-only for the main pipeline. R scripts and R smoke tests have been removed.
amalgkit config has been removed. Selection rules now live in one select_rules.tsv file.
amalgkit curate has been replaced by the split workflow wsfilter -> csfilter -> finalize.
amalgkit csca has been replaced by current cross-species filtering and finalization outputs.
amalgkit rerun was added to rerun failed targets from sanity_report.json.

amalgkit dataset --rule_set base|test|plantae|vertebrate exports bundled select_rules.tsv.
amalgkit dataset --name init creates an empty workspace scaffold.
amalgkit select supports species-wise batch mode and writes summary, queue, and manifest tables.

getfastq supports NCBI, AWS, GCP, ENA, and DDBJ provider fallbacks.
Shared download locks and provider concurrency caps help large parallel runs.
Optional MMseqs2 rRNA and contaminant filters can be ordered with --filter_order.
quant can auto-select kallisto for short-read data and oarfish for long-read data.
Quant index builds use shared locks to avoid duplicate builds under array jobs.

wsfilter performs within-species filtering and writes filtered metadata plus PDFs.
csfilter performs cross-species filtering with BUSCO or orthogroup inputs.
finalize exports final per-species expression tables and supports Python backends for no, sva, ruvseq, combatseq, and latent_glm.

sanity checks getfastq, index, quant, merge, busco, and finalize outputs.
rerun resolves missing or failed targets from sanity reports and writes rerun_manifest.json.

Old command	Current replacement
`amalgkit config`	`amalgkit dataset --rule_set ...` and `select_rules.tsv`
`amalgkit curate`	`amalgkit wsfilter`, `amalgkit csfilter`, `amalgkit finalize`
`amalgkit csca`	`amalgkit csfilter` and `amalgkit finalize`