feat(btrblocks): pick per-chunk bit widths when fastlanes.bitpacked_v2 is allowed - #9754

Summary

Stacked on #9750. Lets the writer emit fastlanes.bitpacked_v2 when its editions permit it. A compressor scheme looks at what is enabled and produces the newest available format.

• CascadingCompressor::with_allowed_serialized_ids / allows_serialized_id. The compressor carries the serialized IDs its output may use. The file writer passes the same set it gives the array context, right where it already filters schemes by allowed encodings, and the compressor knows nothing about editions themselves. Without a restriction, which is what BtrBlocksCompressor::default() gives in-memory callers, every ID is allowed and the newest format wins.
• BitPackingScheme produces the newest allowed format. It picks a width per 1024-element chunk when fastlanes.bitpacked_v2 is allowed and one global width otherwise, so an old edition keeps writing the original format and a newer one gets per-chunk widths from the same scheme. There is no second scheme and no per-scheme wire-ID declaration. The width table child is re-encoded through compress_child, like the children of the decimal and temporal schemes.
• Mostly-patched arrays stay primitive. The scheme returns the original array when half or more of the values would be patches. A few wide values otherwise pack at width 0 with nearly everything patched, which beats raw storage by a couple of buffer bytes and loses them back in footer metadata.
• The CUDA preset allows only each encoding's original format, whose ID is the encoding's own, since CUDA has no per-chunk bit-unpacking kernel yet.
• No shipped edition permits the format yet. A preview entry signals near-certainty about the wire format, and the format was still moving this week. The edition entry can follow once it is blessed; the writer test declares a test-local edition instead.

Tests

• core_writer_never_emits_bitpacked_v2: a session with the default encodings and only the core edition enabled writes a column whose chunks need 1 to 22 bits, and every array it produces reads back as fastlanes.bitpacked.
• permitting_writer_emits_bitpacked_v2: the same session plus a test-local edition permitting the format writes the column as fastlanes.bitpacked_v2 and it reads back equal.
• cuda_compatible_disallows_per_chunk_bitpacking: the default compressor allows the format and the CUDA preset does not.
• test_mostly_patched_stays_primitive: six i64 values, five of them wide, stay primitive instead of becoming width 0 with five patches.
• The new_array_context tests cover the allowed set: the editions' serialized IDs, or every registered ID when editions are disabled.
• The vortex-btrblocks golden snapshots show per-chunk widths and a width_table child for bit-packed columns in both the default and unstable configurations; the four default goldens that change all shrink.

Validation

• cargo nextest run -p vortex-compressor -p vortex-file -p vortex-edition -p vortex-btrblocks -p vortex editions, with and without --features unstable_encodings
• cargo clippy --all-targets --all-features on vortex-compressor, vortex-file, vortex-edition, vortex-btrblocks, and vortex with unstable_encodings; cargo +nightly fmt --all