feat(shadow_metrics): Remove shadow metrics emission

[wbj-cb]

Summary

Dismantles the shadow-metrics background pipeline, leaving the crate as the read-only block-explorer API over persisted shadow blocks. Removes all Prometheus metric emission and the polling reader together with its cursor infrastructure (the reader existed only to poll shadow_blocks via a cursor and emit metrics).

Changes

• shadow-metrics: Delete ShadowMetrics and all metric recording; delete ShadowMetricsReader. /readyz now only checks that shadow_blocks is readable. Drop the base-metrics, metrics, and tokio deps and the metrics feature.
• shadow-indexer-db: Drop ShadowBlockCursor, ShadowMetricsCursorRepo, ShadowBlockRepo::{list_reorged_since, max_cursor}, and ShadowBlockRow::cursor().
• bin: Drop reader spawning, liveness tracking, and poll-interval/max-rows args. The Prometheus metrics initializer is left in place.
• System tests: Keep the DB reconciliation tests (renamed to shadow_blocks_reconciliation.rs), remove the reader/cursor tests, drop the unused base-shadow-metrics dev-dependency.

Notes

• The shadow_metrics_cursor table and its migrations are intentionally left in place; no code references them.

Testing

• cargo clippy clean across base-shadow-indexer-db, base-shadow-metrics, base-shadow-metrics-bin
• Unit tests pass; shadow_blocks_reconciliation system test compiles

[cb-heimdall]

🟡 Heimdall Review Status

Requirement	Status	More Info
Reviews	🟡 0/1	Denominator calculation Show calculation 1 if user is bot 0 1 if user is external 0 2 if repo is sensitive 0 From .codeflow.yml 1 Additional review requirements Show calculation Max 0 0 From CODEOWNERS 0 Global minimum 0 Max 1 1 1 if commit is unverified 0 Sum 1

[depot-code-access]

✅ All benchmarks green — 14 within ±2% (deterministic instruction counts). View run

Benchmark details (14)

Benchmark	Base (target)	Head (this PR)	Δ instructions
batch_queue/drain/drain_cached_span_batches	242,027	242,027	+0.0%
batch_transaction/encode_in_place/encode_in_place	4,199,759	4,199,759	+0.0%
batch_transaction/temporary_frame_buffers/temporary_frame_buffers	8,408,350	8,408,350	+0.0%
flashblock_decode/decode/brotli	3,296,484	3,296,484	+0.0%
flashblock_decode/decode/plain_json	2,280,194	2,280,194	+0.0%
flz/compress_len/real_contract_call	43,148	43,148	+0.0%
flz/compress_len/synthetic_0	38,205	38,205	+0.0%
flz/compress_len/synthetic_1	54,682	54,682	+0.0%
flz/compress_len/synthetic_2	147,976	147,976	+0.0%
flz/data_gas	43,059	43,059	+0.0%
flz/tx_estimated_size	43,056	43,056	+0.0%
frame_parse/decode/single_4kib	1,031	1,031	+0.0%
frame_parse/parse_frames/few_large	1,053,062	1,053,062	+0.0%
frame_parse/parse_frames/many_small	154,763	154,763	+0.0%

[github-actions]

Review Summary

Clean removal of the shadow-metrics polling reader and Prometheus emission pipeline. The deletion is thorough: types (ShadowBlockCursor, ShadowMetricsCursorRepo, ShadowMetricsReader, ShadowMetrics), their re-exports, Cargo dependencies, and the reader system tests are all removed consistently. No dangling references to the deleted types remain in code. The retained reconciliation tests were correctly extracted into a standalone file. The check_schema_ready simplification (dropping cursor-table privilege checks) is appropriate given the cursor infra is gone.

Findings (nits — all in unchanged lines, so inline comments could not be posted):

1. Stale "reader" references in doc comments — Several doc comments in files touched by this PR still reference "the reader" which no longer exists:

   • crates/execution/shadow-indexer-db/src/models.rs:21 — "stall the reader"
   • crates/execution/shadow-indexer-db/src/repo.rs:74 — "keeps a reader from observing"
   • crates/execution/shadow-indexer-db/src/repo.rs:214 — "A row is never emitted until…" (emission was the reader's job)
   • crates/execution/shadow-indexer-db/src/repo.rs:220 — "not the reader's"

   These are not blocking but would be good to clean up in a follow-up to avoid confusing future readers of the code.

No correctness, safety, or concurrency issues found. Not block-production-sensitive (shadow-metrics is a read-only observability service).

[depot-code-access]

Base Std historical fork tests

Fork	Result	Passed	Failed	Skipped	base/base	base-anvil	base-std
Beryl	pass	616	0	13	1e8bec9e	cecddfa5	4658f1b7
Cobalt	pass	721	0	14	1e8bec9e	fb00db40	e30b3421

View run

[github-actions]

Review Summary

Clean removal PR. The deleted code (ShadowMetricsReader, ShadowMetrics, ShadowBlockCursor, ShadowMetricsCursorRepo) is fully unreferenced after the change — grep confirms no remaining call sites. Migration 0008_drop_metrics_cursor.sql correctly uses DROP TABLE IF EXISTS for safe idempotent application.

Minor finding (not in diff, so no inline comment possible)

Stale doc comment in crates/execution/shadow-indexer-db/src/models.rs:21: The ShadowBlockRow.payload field doc still says "Incompatible serde changes fail the whole fetch and stall the reader" — the reader no longer exists after this PR. Consider updating to something like "Incompatible serde changes fail the whole fetch; this risk is accepted."

PR description inconsistency

The PR description says "The shadow_metrics_cursor table and its migrations are intentionally left in place; no code references them." but the PR actually adds migration 0008_drop_metrics_cursor.sql to drop the table. The description should be updated to match the implementation.

No correctness, safety, concurrency, or performance issues found. This PR does not touch block-production-sensitive paths.