Tracking issue for shared source

[xxchan]

RFC risingwavelabs/rfcs#72

• feat(stream): add kafka backfill executor #14172
• feat: add kafka backfill frontend #15602
• CREATE TABLE (no change)
• Blocking DDL for source backfill #18338
• Let source backfill finish when there's no data from upstream #18299
• pause/start from latest to avoid waste work
  • pause SourceExecutor before a Backfill is created. This means the SourceExecutor doesn't waste work, and the first MV (maybe we even want multiple MVs?) doesn't need backfill. feat: pause shared SourceExecutor until a downstream actor is created #16348
  • shared source: don't backfill for the first MV #16576
  • feat(source): let shared SourceExecutor start from latest instead of specified offset #16626
• Check whether metrics need refinement (whether cloud needs change)
• feat: enable shared source in session variable by default, and add cluster-level config to disable #18749
• benchmarks https://linear.app/risingwave-labs/issue/PERF-163/[test-request]-disable-shared-source-by-default-in-the-daily-benchmark

ALTER

• feat: ban ALTER SOURCE for shared source, and refine all ALTER clauses #18750
• schema change (Tracking: Alter Source/table with connector #11800, Tracking: schema changes of source | (table with connector)  #15582)
  • Support ALTER SOURCE ADD COLUMN for shared source #19063
• scaling
  • feat: support scale SourceBackfill executor #16825
  • Support ALTER SOURCE SET PARALLELISM #19508
  • Problem of online scaling for source backfill #18300
• Support ALTER BACKFILL_RATE_LMIT for SourceBackfill #19297

testing

• more e2e tests improve script for running e2e source tests #15995
• test: add some test for split change #15994
• backfill test test: add kafka backfill large data test #18924