Fix SS migration doc

[yzang2019]

Describe your changes and provide context

Fix some minor outdated text for SS migration guide.

Testing performed to validate your change

[cursor]

PR Summary

Low Risk
Documentation-only changes to the migration guide; no runtime or config code is modified.

Overview
Updates docs/migration/giga_store_migration.md so operators see current on-disk layout and migration options, not the older data/evm_ss/ wording.

Paths and legacy behavior: Cosmos SS is documented as data/state_store/cosmos/{backend} and EVM SS as data/state_store/evm/{backend}, with notes that existing data/pebbledb/ or data/evm_ss/ trees are still used when already present.

Config and verification: The app.toml example now includes evm-ss-separate-dbs = false (recommended single-DB EVM layout). The verification section adds sample startup log lines and clarifies that separateDBs=false does not mean Giga SS is disabled—it only reflects the experimental per-type split flag.

Migration paths: Prerequisites and Step 2 distinguish P2P state-sync (layout-agnostic on the wire) from layout-sensitive data/ tarballs, including when a Giga snapshot restore can skip another P2P sync. Rollback, safety checks, and FAQ entries are aligned with the new paths and the two new FAQ answers on logs and tarball migration.

^{Reviewed by Cursor Bugbot for commit 58ef462. Bugbot is set up for automated code reviews on this repo. Configure here.}

[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	Aug 27, 2026, 4:04 PM

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 60.24%. Comparing base (d69e67d) to head (58ef462).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4027      +/-   ##
==========================================
- Coverage   61.24%   60.24%   -1.00%     
==========================================
  Files        2153     2054      -99     
  Lines      188393   176643   -11750     
==========================================
- Hits       115373   106419    -8954     
+ Misses      62289    60462    -1827     
+ Partials    10731     9762     -969     

Flag	Coverage Δ
sei-db	69.80% <ø> (ø)
sei-db-state-db	?

Flags with carried forward coverage won't be shown. Click here to find out more.
see 142 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.

[seidroid]

Docs-only update to the Giga SS migration guide; the new path layout (data/state_store/{cosmos,evm}/{backend}), the legacy data/evm_ss fallback, the evm-ss-separate-dbs key and the sample startup log lines all check out against sei-db/common/utils/path.go, app/seidb.go and sei-db/state_db/ss/composite/store.go. One gap: the new "restore a data-directory snapshot" path omits the config that must match the snapshot.

Findings: 0 blocking | 4 non-blocking | 1 posted inline

Blockers

• None at the file/PR level.

Non-blocking

• None at the file/PR level.
• 1 suggestion(s)/nit(s) flagged inline on specific lines.
• 3 non-blocking pre-existing issue(s) listed below under pre-existing issues.

Pre-existing issues

• [suggestion] docs/migration/giga_store_migration.md "What's different about EVM SS" (lines 32-36, unchanged by this PR) says EVM SS is "point-query only by design" and "iteration is explicitly disabled on the EVM backend". The code supports it: CompositeStateStore.Iterator/ReverseIterator route EVM store keys to EVMStateStore, which serves them from the unified DB when separateDBs=false (sei-db/state_db/ss/evm/store.go:121). The section's rationale ("cross-bucket scans would defeat the per-type sub-DB layout") only applies to the evm-ss-separate-dbs = true mode this PR now documents as non-default and experimental, so the section reads as more restrictive than reality.
• [suggestion] sei-db/config/toml.go:157 still documents evm-ss-db-directory as defaulting to <home>/data/evm_ss, while sei-db/config/ss_config.go:124 and utils.GetEVMStateStorePath use <home>/data/state_store/evm/{backend} (with data/evm_ss only as a legacy fallback). This PR fixes exactly that staleness in the migration guide; the generated app.toml comment operators actually read has the same issue.
• [suggestion] The "Rollback Steps" section (lines 189-198) lists only the config flip and directory cleanup, without pointing at the FAQ entry (lines 222-227) that explains EVM state written while split was enabled is absent from Cosmos SS, so historical EVM queries stay broken until another state sync. An operator following Rollback Steps alone will not see that caveat.

[seidroid]

[suggestion] Restoring a Giga data/ tarball needs more than evm-ss-split = true: the restoring node's ss-backend (and ss-db-directory / evm-ss-db-directory if set) must match the layout inside the tarball, because both SS paths are backend-qualified (utils.GetStateStorePath / GetEVMStateStorePath → data/state_store/{cosmos,evm}/{backend}).

The backend mismatch is the dangerous case and it fails silently: a pebbledb tarball restored on a node configured with ss-backend = "rocksdb" resolves both SS dirs to non-existent .../rocksdb paths, so Cosmos SS opens empty — and because the safety checks in NewCompositeStateStore only fire when "Cosmos SS already has history", none of them trip. The node starts against a restored SC with an empty SS instead of refusing to launch. (A mismatched evm-ss-separate-dbs is caught: the EVM dir is non-empty but its unified DB is, so validateEVMSSPreRecovery aborts startup.)

Suggest adding a line here — and to the matching FAQ answer at line 243 — that the restoring node's ss-backend, evm-ss-separate-dbs, and any SS directory overrides must match the node the snapshot was taken from.