fix(fullnode): avoid rewriting pending proofs on retry

[tonykim525]

Description

This PR fixes unnecessary disk-write amplification in the PendingProof retry flow.

When a proof remains pending, retry processing could re-store the same pending proof every L1 cycle. This increases write pressure/disk usage without changing logical state.

This PR includes:

• a precheck reproduction test (sov-db)
• the fullnode retry-path fix (da_block_handler.rs)

Implementation summary:

• Added ProofSource (FromL1, FromPendingRetry).
• Fresh L1 proofs use FromL1; pending-table retries use FromPendingRetry.
• store_pending_proof(...) now runs only for FromL1.
• Retry-origin pending proofs are kept pending without rewrite.
• Updated logs to match actual behavior.

Logical semantics are preserved: while a proof is still pending, its existing PendingProof entry is kept without rewrite across retry cycles.

Testing

Targeted reproduction:

• cargo test -p sov-db --features "sov-schema-db/test-utils" test_pending_proofs_current_flow_rewrite_increases_disk_usage -- --nocapture

Observed:

• current_rewrites=40, fixed_rewrites=0
• current_pre_flush_db_bytes=2981196, fixed_pre_flush_db_bytes=357396
• current_post_flush_db_bytes=428461, fixed_post_flush_db_bytes=362817
• current_live_bytes=66616, fixed_live_bytes=66616
• current_total_sst_bytes=66616, fixed_total_sst_bytes=66616
• Result: 1 passed; 0 failed; 7 filtered out (targeted test-name filter)

Interpretation:

• Current flow rewrites once per cycle; fixed flow skips rewrites.
• Pre-flush disk footprint is significantly higher in current flow (~8.34x).
• Live data/SST converge in this synthetic setup, so logical state is preserved while redundant writes are removed.

Full sov-db package test run:

• cargo test -p sov-db --features "sov-schema-db/test-utils" -- --nocapture
• Result: 8 passed; 0 failed (doc-tests: 0 passed; 0 failed; 2 ignored)

Additional validation:

• rustfmt +nightly --check crates/fullnode/src/da_block_handler.rs
• cargo check -p citrea-fullnode (passed)
• cargo test -p citrea-fullnode --no-run (passed)
• cargo test -p citrea --features testing --test bitcoin_e2e --no-run (passed)

Docs

No public API/interface changes.
No docs update required.

[codecov]

Codecov Report

❌ Patch coverage is 80.64516% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.5%. Comparing base (a030fa8) to head (2838b74).

Files with missing lines	Patch %	Lines
crates/fullnode/src/da_block_handler.rs	80.6%	6 Missing ⚠️

Additional details and impacted files

Files with missing lines	Coverage Δ
crates/fullnode/src/da_block_handler.rs	86.0% <80.6%> (-0.2%)	⬇️

... and 3 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[tonykim525]

test(sov-db): make pending proof repro test clippy-compliant - 0d950d3

Only Commands run:

1. rustfmt +nightly crates/sovereign-sdk/full-node/db/sov-db/src/ledger_db/tests.rs
2. cargo clippy -p sov-db --all-targets --all-features --features "sov-schema-db/test-utils"
3. SKIP_GUEST_BUILD=1 make lint

[tonykim525]

I can see the Validate PR Title check is failing on this PR.

It did pass once earlier, and now appears to be failing due to workflow-side behavior.
From my side, this does not look fixable in this fork.
Could the maintainers please take a look and help unblock this check?

I addressed the two previous CI issues in follow-up commits:

• fixed lint/format and clippy issues
• added the required changelog entry

Could you please re-run CI on the latest commit?
If there’s anything else you’d like me to adjust, I’m happy to update it (tests, coverage, etc.).

[jfldde]

Hey @tonykim525,
Just approved to re-run on latest commit. Thanks for the PR, it looks great. I'll give it a review tomorrow and try it out. Don't worry about the Validate PR Title for now, I'll open an issue regarding this.
Regarding #3168, I was more thinking about reducing PendinfProofs disk usage by storing bitcoin (block idx, tx idx) instead of storing the proof itself and going through a RPC call on re-processing.

[jfldde]

Great PR. Removed the Fixes #3168 as this doesn't directly address it.

[tonykim525]

Sorry — I moved the changelog entry before merging nightly. I’ve now resolved the changelog conflict and placed it under Unreleased as requested.

Removed the pending-proof disk-usage repro test, moved the changelog entry under Unreleased, and resolved the nightly merge conflict.

Validated with:

• cargo test -p sov-db ledger_db::tests --no-run
• rustfmt +nightly --check /Users/kim-yewon/IdeaProjects/citrea/crates/sovereign-sdk/full-node/db/sov-db/src/ledger_db/tests.rs

Both passed.