WIP: performance optimizations

[KRRT7]

Description

Performance optimization stack focused on terminal hot paths:

• Optimizes terminal flat-storage rebuild/replay paths, including arithmetic fast paths, capacity reuse, dense-row replay, cursor resize hardening, differential/property tests, fuzz coverage, and Criterion benchmark coverage.
• Reduces fuzzy/file search and input-edit hot-path work, including safer multibyte handling in AI context edits.
• Avoids redundant terminal block-list height/layout work for rich content updates.

Latest VM benchmark summary after cargo clean, comparing upstream/master b43821d0 against branch a2c2a9e1 with representative flat-storage rebuild workloads:

• Single resize cases were 64.7%-89.1% faster across command history, cargo build logs, long soft-wrapped output, agent transcripts, and cursor blank-tail sessions.
• Resize sequence cases were 68.4%-85.5% faster.
• Largest representative win: cargo-build-log single resize to 160 cols, 1.65ms -> 0.18ms, 89.1% faster.
• All measured cases improved with Criterion significance p < 0.05.

Workload	Single resize improvement	Resize sequence improvement
command history	65.8%-72.7% faster	70.7% faster
cargo build log	84.0%-89.1% faster	85.5% faster
long soft-wrapped output	72.4%-81.2% faster	76.9% faster
agent transcript	66.7%-76.4% faster	72.5% faster
cursor blank-tail session	64.7%-76.2% faster	68.4% faster

Linked Issue

• The linked issue is labeled ready-to-spec or ready-to-implement.
• Where appropriate, screenshots or a short video of the implementation are included below (especially for user-visible or UI changes).

Testing

• cargo fmt --all
• env -u RUSTC_WRAPPER cargo nextest run -p warp -E 'test(test_gap_after_banner) | test(test_removed_gap_with_banner) | test(test_update_padding_block_heights) | test(test_remove_rich_content_block) | test(test_update_rich_content_heights_in_lines_updates_changed_items_by_position)'
• git diff --check upstream/master..HEAD
• VM Criterion comparison: temporary representative flat_storage rebuild harness, upstream/master b43821d0 vs branch a2c2a9e1; artifacts in /root/warp/target/criterion/flat_storage_representative_*.

Not yet run: full ./script/presubmit / full workspace clippy.

• I have manually tested my changes locally with ./script/run

Screenshots / Videos

Not applicable: internal performance changes with benchmark coverage.

Agent Mode

• Warp Agent Mode - This PR was created via Warp's AI Agent Mode

CHANGELOG-NONE