[skip ci] v2.17.0 release announcement (#4653)

website/release_announcement_drafts/v2.17.0.md

@@ -0,0 +1,28 @@
+This release adds
+
+- Improved and bugfixed rendering of modified bases from BAM/CRAM files,
+  particularly, it matches in many ways the IGV rendering of modified bases [^1]
+- Improved performance for ultra-long reads from Nanopore, especially with
+  zooming in and out and scrolling because it caches the mismatches instead of
+  re-calculating them repeatedly [^2]
+- Improved performance for wiggle tracks, multi-wiggle tracks, and SNP coverage
+  tracks, particularly changing the height for these track types produces less
+  re-rendering
+- A new data adapter loading BLAST tabular -outfmt 6 alignments for the synteny
+  view
+- Fixed interpretation of the 'ts' tag from minimap2 for RNA-seq strandedness
+
+[^1] Importantly, JBrowse 2 now takes into account the ML tag (likelihood) to
+better report the results when tools are providing low probability modification
+likelihoods, especially for when there are multiple modification possibilities
+at each position (e.g. commonly outputted from Dorado which reported as 'm' and
+'h' probabilities, we now only choose the highest likelihood for reporting,
+which is in line with interpretation from
+[the MM/ML tag specification](https://samtools.github.io/hts-specs/SAMtags.pdf))
+
+[^2] As part of this change, the internals of BAM/CRAM records have been
+changed. If you have code that was directly accessing any internal attributes of
+these records, it may not work anymore. Please use the feature.get('...') on
+features to get data of interest. The .get(...) method is the only "official"
+API for accessing data on feature objects, and contact us if you run into any
+problems!