Review

This PR adds a single documentation file (docs/changes/20260720-alchemy-bundler-provider.md, +36/-0) — no Go or Solidity source changes. Its stated purpose is procedural: trigger release-on-pr-close.yml against main so semantic-release picks up the feat: bundler_provider commit (b061b1c) that landed via a direct rebase+push (which bypassed the PR-merge event).

Since there's no code diff, the usual Go checklist (concurrency safety, error handling, test coverage, etc.) doesn't apply here. I did fact-check the doc's technical claims against the code currently on main, since accuracy is the whole point of a change-doc:

• ✅ SmartWalletConfig.BundlerProvider / AlchemyAPIKey fields, ActiveBundlerURL(), and BundlerConfigured() all exist as described in core/config/config.go.
• ✅ ActiveBundlerURL() fails closed for the alchemy path (missing key / unmapped chain → hard error, no silent fallback to bundler_url) — matches lines 259–269.
• ✅ alchemyNetworkSubdomain map matches the doc exactly: 1→eth-mainnet, 11155111→eth-sepolia, 8453→base-mainnet, 84532→base-sepolia, 56→bnb-mainnet.
• ✅ eth_simulateUserOperation best-effort fallback confirmed in pkg/erc4337/preset/builder.go:818-829 (warns and continues rather than failing hard on unsupported bundlers).

One minor nit: the PR title is docs: change-doc for Alchemy bundler rollout (cuts v4.4.0 pre-release). Per CLAUDE.md's Conventional Commits requirement, docs: is a non-version-bumping type under standard semantic-release config — worth double-checking that go-semantic-release's config here actually cuts a release off a docs: PR title, or whether the version bump instead comes purely from the already-on-main feat: bundler_provider commit being included in the release computation (which the PR body implies). If it's the latter, the PR title itself doesn't need to carry release-triggering weight, so docs: is fine and this is a non-issue — just flagging it since the body's framing ("cuts v4.4.0 pre-release") could be read as depending on this PR's own title.

Nothing else to flag — the doc is accurate, clearly written, and the change is low-risk (docs-only, targets main per the standing exception for release-cut PRs).