multi: add labels for static addresses

[GustavoStingelin]

Adds persistent, locally managed labels for static addresses.

• Stores labels in LoopDB
• Exposes labels through RPC responses and CLI commands
• Supports updating and clearing labels
• Associates labels with the address and UTXOs

[gemini-code-assist]

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

[GustavoStingelin]

Note: reviewers, pls, pay attention here.

[starius]

I propose to include the current state of #1139 into this PR, so own label commits go after @hieblmi's commits. Add a note to the PR description not to review the multi-address commits.

I attempted to do a rebase using AI, here is my attempt: https://github.com/starius/loop/commits/staticaddr-combined/ It is not functional.

Some semantic rebase is needed. For example:

• PR 1178 adds 000022_static_address_label, while PR 1139 already uses migrations 22–24 for multi-address persistence.
• PR 1178 assigns protobuf field 2 in NewStaticAddressRequest to label; PR 1139 assigns field 2 to send_coins.
• PR 1178 assigns protobuf field 3 in NewStaticAddressResponse to label; PR 1139 assigns field 3 to send_coins_response.

Found some issues in that multi-address + labels branch that I got:

• the multi-address manager returns an internal *Parameters pointer after releasing its lock, while label update mutates that same object. go test -race reports a real concurrent read/write race. repro
• after multi-address is applied, static new returns a derived receive address, but static summary still reads the legacy/root seed. The two commands no longer describe the same address. repro
• every generated change address is stored with an empty label, regardless of the labels on the consumed deposits. I think labels should be inherited by change - that is a common practice among wallets
• testing gap: the SQL tests read the label back through the same open database/store instance. They do not close and reopen the database or rebuild the manager cache.
• CLI: the shared --label help text says "label for this swap" and lacks a space, while static deposit --label labels the lnd funding transaction and static new --label labels the address.
• ListUnspentDeposits includes the label, but historical deposit/list responses carry the address without the corresponding label - needs unification

Some of them will be resolved by a proper rebase, but some seem genuine.

[GustavoStingelin]

Okay, if we agree to rebase this on top of #1139, I’d prefer to wait until @hieblmi has finished it and its CI is green. This should avoid premature conflicts and duplicated work from having to address #1139’s current race failures in the combined branch.