[marcoarnulfo]

Fixes #1283.

do_sort always wrote to wb.active_sheet, and data_range dropped the Sheet! prefix of range, so data.sort (and sort_range, which goes through it) sorted the same cells on the active sheet and reported success. Sort now resolves its sheet with the existing target_sheet helper, the way range.setValues and the formatting commands do, and passes it down to do_sort; data.sortAscending/data.sortDescending share that path, so they get the same fix. Following the rule in #885, the selection belongs to the active sheet, so naming another sheet without a range is an error, and so is a prefix that disagrees with sheet. Happy to switch to refusing a prefix that names another sheet instead (the #206 rule for cell references), if you'd rather keep sort on the active sheet.

Thanks @Essermarth for the precise repro. Other data.rs commands still read active_sheet directly (filter, remove duplicates, subtotal, circle invalid data); I left them out to keep this one small, happy to send them as a follow-up.

Validation:

• tests::sort_with_a_sheet_prefix_sorts_the_named_sheet (the steps from the issue): fails on main (Prices stays p1, p2, p3, Sheet1 gets sorted), passes here
• cargo test -p gridcraft-engine --lib sort, cargo clippy -p gridcraft-engine --all-targets -- -D warnings, cargo fmt --all -- --check
• (macOS, Apple silicon)

[Essermarth]

Thanks for the quick fix! Sorting the named sheet, as your PR does, is exactly the behavior we expected: an agent passing "Prices!A1:C5" means that sheet, so please keep it that way rather than refusing the prefix. A follow-up for the other data.rs commands (filter, remove duplicates, subtotal) would be very welcome too, since an agent hits them the same way. (AI-assisted testing.)

[marcoarnulfo]

Thanks, good to know that's the behaviour you want, I'll keep it. I'll do the follow-up for filter, remove duplicates and subtotal (and the other data.rs commands that read the active sheet) in a separate PR.