fix(BREAKING): upgrade to dprint-core 0.69.1 and remove anyhow

[dsherret]

Upgrades dprint-core from 0.67.0 to 0.69.0 (keeping the process feature).

Why source changes were needed

dprint-core 0.68.0 removed its anyhow dependency, so the plugin handler APIs now use dprint_core::plugins::FormatError instead of anyhow::Error. FormatResult is now Result<Option<Vec<u8>>, FormatError>, and handle_process_stdio_messages returns Result<(), FormatError>.

Changes

• Cargo.toml: dprint-core 0.67.0 -> 0.69.0, dropped the direct anyhow dependency.
• src/handler.rs: replaced anyhow::Error/anyhow::Result/anyhow!/bail! with FormatError and FormatError::new(...). The ? plumbing over JoinError, oneshot::error::RecvError and std::io::Error still works because dprint-core implements From for those on FormatError. SetupInitError::Failed now holds a FormatError.
• src/main.rs: fn main() -> Result<(), FormatError>.
• Cargo.toml / tests/tests.rs: dprint-development 0.10.1 -> 0.11.0. This was forced — 0.10.x's run_specs callback signature is anyhow::Result<Option<String>>, which the plugin can no longer produce without keeping anyhow purely for the test harness. 0.11.0 uses Result<Option<String>, Box<dyn std::error::Error + Send + Sync>>, so the spec test just adds .map_err(Into::into).

No behavior changes: all error messages are byte-for-byte identical to before.

dprint_core::formatting::ir_helpers::GeneratedValue gained an is_known_multi_line field in 0.69.0, but this plugin doesn't use the IR, so nothing to do there.

Verification

• cargo check --all-targets clean
• cargo test: 18 passed, 0 failed (16 unit, 1 spec test covering 9 sub-specs, 1 doc test)
• cargo clippy --all-targets: only the two pre-existing warnings (manual arithmetic check, needlessly owned Cow type)
• Formatted with rustfmt --edition 2024 and dprint fmt