[thomasklemm]

Alternative to #571: also fixes the clock, but near the fixture's dates, so the 16 time-filtered requests render stories. Merge one of the two.

The fixture's stories are from 2023-07-18, so on main /active, /recent and /top?length=1d|1w|1y (16 of 114 requests per iteration) render empty lists. This fixes Time.now at 2023-07-19 01:00 UTC, about 15 hours after the stories, during boot and in the benchmark block. Across the four tested dates (2026-11-05, 2026-12-25, 2027-03-20, 2027-06-28), allocations per iteration varied by 0.003%, against 1.37% on main.

/top never listed a story on main: DATETIME('now', '- 1 DAY') is NULL, because SQLite rejects - 1 and has no WEEK. The modifier is now -1 DAY, with weeks as 7 days, as in upstream lobsters. The deliberate departure from upstream is that the query binds Time.current instead of SQLite's 'now', so /top follows the fixed clock.

Per pass after one warmup pass, Ruby 4.0.9:

	stories in lists	KB	objects
main	145	3,709	658,492
#571	145	3,709	649,706
this PR	545	4,771	950,426

The 16 requests render 25 stories each. Since stories and comments are hours old, pages also take the branches for recent content: reply and flag links, delete instead of disown, comment scores under 36 hours shown as ~, one user marked as new. This is a deliberate workload change of about +44% objects per iteration against main (+46% against #571).

Response bodies are byte-identical between runs on 2026-11-05 and 2027-01-20, apart from CSRF tokens. The harness keeps the real clock. Date.today, DateTime.now and Time.new without arguments stay live. Cache entries and the session no longer age. TRACK_AR_TIME uses the monotonic clock.

How this was measured

Per pass: route_dump.rb records status, bytes, <li id="story_ count and allocated objects of each request in the pass after one warmup pass; route_table.rb sums them. Date spread: GC.stat(:total_allocated_objects) around each iteration of the default harness, WARMUP_ITRS=1 MIN_BENCH_ITRS=3, mean of the 3 measured iterations per date, spread = (max − min) / min of the per-date means, system clock set with libfaketime (FAKETIME='@<date> 12:00:00' FAKETIME_DONT_FAKE_MONOTONIC=1), cg_harness.rb with CG_DUMP=0. Responses: raw_dump.rb writes status, headers and body of all 114 requests in the pass after one warmup pass; raw_cmp.rb blanks the authenticity_token values, the csrf-token meta tag and the random headers (session cookie, ETag, X-Request-Id, X-Runtime) and compares the rest byte for byte: 0 of 114 differ.