Convert sequential single deletes into range tombstones

[joshkang97]

Summary

Add a read-path optimization that converts contiguous point tombstones into range tombstones during forward iteration. When a configurable threshold of consecutive point deletions (with no live keys between them) is detected, a range tombstone covering [first_tombstone_key, next_live_key) is inserted into the active mutable memtable. This benefits future iterators by enabling MergingIterator's cascading seek optimization to skip over large deleted ranges instead of scanning through individual tombstones one by one.

If there is a memtable switch during the read iteration, then the range deletion entry is discarded.

The inserted range tombstones are logically redundant (they don't delete anything that isn't already deleted by point tombstones), skip WAL (they're a derived optimization regenerated by future reads on crash), and use the read snapshot's sequence number so they don't interfere with newer writes.

Key changes

• New option min_tombstones_for_range_conversion (AdvancedColumnFamilyOptions): Threshold of contiguous point tombstones before converting to a range tombstone. Default 0 (disabled). Dynamically changeable via SetOptions().
• DBIter tracking logic (db/db_iter.cc, db/db_iter.h): Tracks contiguous tombstones during FindNextUserEntryInternal(). When a live key (kTypeValue, kTypeMerge, etc.) is encountered after enough tombstones, calls MaybeInsertRangeTombstone(). Tracking is reset on Seek/Prev/direction changes.
• MemTable::AddLogicallyRedundantRangeTombstone() (db/memtable.cc, db/memtable.h): New method on ReadOnlyMemTable interface that inserts a range tombstone using concurrent memtable insert. Returns false if the memtable has already been switched to immutable (fragmented range tombstone list already constructed).
• New statistics tickers: READ_PATH_RANGE_TOMBSTONES_INSERTED and READ_PATH_RANGE_TOMBSTONES_TOSSED to track how many range tombstones are inserted vs skipped.

Test Plan

• New unit tests
• Stress test coverage

Benchmark results

Setup: fillseq,compact 1M entries, then deleteseq,flush 500K entries
Workload: seekrandom, seek_nexts=100, threads=8, duration=10, 3 runs averaged

Branch	ops/s	% change vs main
main	300	—
range_tombstone_insert (threshold=8)	269,389	+89,696%
range_tombstone_insert (threshold=0)	290	-3.3%

No-regression benchmark (compacted DB, no tombstones):

Setup: fillseq,compact 1M entries (no deletes)
Workload: seekrandom, seek_nexts=100, threads=8, duration=10, 3 runs averaged

Branch	ops/s	% change vs main
main	341,378	—
range_tombstone_insert (threshold=8)	342,449	+0.3%
range_tombstone_insert (threshold=0)	344,238	+0.8%

No measurable regression on a clean compacted DB with no tombstones.

[github-actions]

✅ clang-tidy: No findings on changed lines

Completed in 632.6s.