fix(strategies): empty a folder before deleting it on the core

[guyverino]

What & why

Right-click → Delete on a strategies folder that still held (disabled) strategies did nothing: the confirmation closed, the folder stayed on the core. The terminal sent only TStratDelete(0, path), and the core refuses that for a populated folder — its log says Folder delete request … Demo (not empty or not found) and no reply event reaches moonproto, so the UI had nothing to show. Per moonproto docs/strats.md, the legacy command deletes an empty folder only, and omission from the folder tree cannot remove a folder a retained strategy occupies: there is no wire command that removes a folder together with its contents.

StrategiesView::delete_folder now queues a DeleteStrategy for every row under the folder (subfolders included; all confirmed disabled by the existing folder_delete_authorized guard) and then the folder command, in the same per-core FIFO the feed drains in order, so the core reads the folder delete once the folder is empty. Selection state is cleared of the deleted rows as for a plain strategy delete. Three doc comments that claimed the legacy command deletes the contents now state what the core does; a source-grep test pins the row-then-folder ordering.

Notable decisions

• The populated case finishes with the legacy delete(0, path) rather than a folder-tree omission even on cores with a versioned tree: at drain time the local snapshot still holds the rows (no echo yet), and the protocol keeps a folder a retained strategy occupies. The legacy command deletes an existing empty folder on either kind of core, and moonproto applies its echo to the versioned tree.
• Another client adding a row between confirmation and the drain makes the core refuse the folder command — the folder stays, nothing else is lost. That is the accepted failure mode.
• The "enabled strategies inside" rule is unchanged: a folder with any enabled strategy still refuses with the existing notice, and the confirmation is revalidated against the live snapshot on Yes.

Known limitations

• The core still gives no reply event on a refused folder delete, so a refusal remains visible only in the core's own log.
• Spotted, not touched (separate topic): strategies/tree/dnd.rs cut/paste cleans an emptied source folder with delete_empty_folder(…, Vec::new()); the feed guard compares the core's live placements with that empty list, so on any core holding a strategy the command is skipped.

Issues

None of the 16 open issues is touched by this change.

How to verify

cargo fmt --all -- --check
cargo build -p moon-ui-gpui --bin moonterminal --target x86_64-pc-windows-msvc --all-targets
cargo clippy -p moon-ui-gpui --bin moonterminal --target x86_64-pc-windows-msvc --all-targets -- -D warnings
cargo test -p moon-core --target x86_64-pc-windows-msvc
cargo test -p moon-ui-gpui --target x86_64-pc-windows-msvc

fmt check exit 0 · build exit 0 · clippy: no new finding (the 6 locations reported in the two touched moon-core files are the same 6 on a detached origin/main, shifted by the inserted comment lines) · tests: moon-core 1961 + moon-ui-gpui 1816 + 333 passed, 0 failed · FireTest not run (no chart/render/window/input surface).

Live: on a real core, Delete on a folder with disabled strategies → app log shows delete strategy … ×N followed by delete folder <path>, the core log shows Strategy deleted ×N and no (not empty or not found).

🤖 Generated with Claude Code