fix(prisma): stop migrate status panicking on a trailing migration id

[KuSh]

Problem

filter_migrate_status (src/cmds/js/prisma_cmd.rs) finds the 202… migration id, looks for the whitespace that ends it, and falls back to a hardcoded 20 bytes when there is none:

let end = line[pos..].find(|c: char| c.is_whitespace()).unwrap_or(20);
latest_migration = line[pos..pos + end].to_string();

When the id runs to the end of the line there is no whitespace, so the slice runs past it:

Migration could not be applied at 2024-01-01T12:00:00
len 53, pos 34, end 20  ->  line[34..54]
thread panicked: end byte index 54 is out of bounds

Release builds set panic = "abort" (Cargo.toml), so rtk prisma migrate status dies outright and the user loses the command's output — the failure mode .claude/rules/rust-patterns.md singles out ("If filter fails, execute raw command unchanged. Never block the user").

A multibyte character straddling pos + 20 panics the same way, as a char-boundary error instead.

Fix

Fall back to the rest of the line, which is what the sibling filter_migrate_dev has always done 70 lines above:

let end = line[pos..]
    .find(|c: char| c.is_whitespace())
    .unwrap_or(line.len() - pos);

That bound is char-safe by construction: line.len() - pos is exactly the length of the line[pos..] slice.

Test plan

filter_migrate_status had no test at all (only test_filter_migrate_dev existed). Added one covering both shapes — an id at end of line, and a multibyte character right after it — and verified it panics at prisma_cmd.rs:320 with the old fallback restored, then passes with the fix.

cargo fmt --all && cargo clippy --all-targets && cargo test --all clean.

Found by a review round while working on another PR; unrelated to that work, so it is on its own branch off develop.

🤖 Generated with Claude Code

[aeppling]

lgtm