autobahn: add data_prune_after to bound data.State memory (CON-256)

[wen-coding]

data.State.runPruning is a background goroutine that drops in-memory blocks/QCs/AppProposals older than a configurable duration, but the config knob (data.Config.PruneAfter) was never wired up — giga_router constructed data.NewState with only Committee set, so the pruner never spawned. data.State.PruneBefore (the giga_router-driven path based on cosmos-sdk RetainHeight) is also a no-op when the chain is configured with pruning="nothing" (the localnode default, common in test setups), so in-memory data.State grew with the chain under sustained load and eventually OOM-killed nodes.

Plumb DataPruneAfter through:
`AutobahnFileConfig.data_prune_after` (json) → `GigaRouterConfig.DataPruneAfter` → `data.Config.PruneAfter` → `data.State.runPruning`.

Production default (gen-autobahn-config): 30m, gives operators plenty of recent history for /block, /tx, /trace_*, etc. while bounding memory under load. Localnode/test override (step4_config_override.sh): 1m, keeps data.State small under sustained-throughput tests where cosmos pruning is "nothing".

Things done

• AutobahnFileConfig.DataPruneAfter
• GigaRouterConfig.DataPruneAfter + thread into data.Config
• node/setup.go pass-through from autobahn.json
• gen-autobahn-config production default (30m)
• step4_config_override.sh localnode override (1m)
• gofmt, vet clean

[github-actions]

The latest Buf updates on your PR. Results from workflow Buf / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
✅ passed	✅ passed	✅ passed	✅ passed	May 4, 2026, 4:13 AM

[codecov]

Codecov Report

❌ Patch coverage is 66.66667% with 3 lines in your changes missing coverage. Please review.
✅ Project coverage is 59.17%. Comparing base (6620fb6) to head (f3cf900).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
sei-tendermint/node/setup.go	50.00%	1 Missing and 1 partial ⚠️
...int/cmd/tendermint/commands/gen_autobahn_config.go	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3375      +/-   ##
==========================================
- Coverage   59.17%   59.17%   -0.01%     
==========================================
  Files        2097     2097              
  Lines      172641   172648       +7     
==========================================
+ Hits       102163   102167       +4     
- Misses      61615    61617       +2     
- Partials     8863     8864       +1     

Flag	Coverage Δ
sei-chain-pr	68.98% <66.66%> (?)
sei-db	70.41% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
sei-tendermint/config/autobahn.go	28.57% <ø> (ø)
sei-tendermint/internal/p2p/giga_router.go	69.30% <100.00%> (+0.46%)	⬆️
...int/cmd/tendermint/commands/gen_autobahn_config.go	18.18% <0.00%> (-0.34%)	⬇️
sei-tendermint/node/setup.go	69.23% <50.00%> (-0.35%)	⬇️

🚀 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.

[pompon0]

fyi, pruneAfter in data was used in sei-v3, but in sei-chain it is application which is responsible for pruning, via retainHeight field in ResponseCommit. I currently don't know if we want to change ownership of pruning from application to consensus. IMO it would make sense, given that application should rather be solely concerned with the latest state at all times. However perhaps sei-chain app makes some assumptions wrt which blocks are available (I can imagine that it does, but I haven't looked into that yet).

[wen-coding]

fyi, pruneAfter in data was used in sei-v3, but in sei-chain it is application which is responsible for pruning, via retainHeight field in ResponseCommit. I currently don't know if we want to change ownership of pruning from application to consensus. IMO it would make sense, given that application should rather be solely concerned with the latest state at all times. However perhaps sei-chain app makes some assumptions wrt which blocks are available (I can imagine that it does, but I haven't looked into that yet).

Sorry I'm confused. Not planning to change prune ownership in this PR (although we can discuss whether that should be done, I'm generally of the opinion this is consensus cleanup which should probably be controlled via consensus), I just want to set a smaller prune period in tests, so that in less powerful machines (my Mac) we can still run long throughput tests without the validators getting OOM. The 30m default in gen-autobahn-config is a defensive cap (still opt-out — operators can drop the field), not a replacement for app-driven pruning.

[pompon0]

Currently pruning is driven by retainHeight computed via

sei-chain/sei-cosmos/baseapp/abci.go

Line 680 in 52d368f

func (app *BaseApp) GetBlockRetentionHeight(commitHeight int64) (int64, error) {

. If pruning in tests does not work, we should first check whether this function actually advances retainHeight and fix it if it doesn't (or if it doesn't keep up with the block rate).