fix(terminal_view): 修复 SSH 连接窗口空行的时间戳占位括号与行号

[feigeCode]

Closes #244

Description

The per-line margin (timestamps + line numbers, introduced for #220) drew a full-width row for every on-screen line, so lines that had not produced any output yet still got a [ ] timestamp placeholder — and later, an orphan line number under the cursor. In a fresh SSH session that littered the whole window below the cursor.

This PR makes “has this line produced output yet?” the single criterion for the entire margin: a margin row is only rendered for lines that already have a timestamp in the line timeline. Rows without a label become an empty string, and the element already skips empty rows — so neither the placeholder brackets nor the line number is painted.

• crates/terminal/src/line_timeline.rs: add SharedLineTimeline::labels(first_id, count) — one lock for the whole screen, O(log n + count), and one consistent per-frame snapshot. The single-row label() had no production caller left after the switch, so it was removed rather than kept as a parallel path.
• crates/terminal_view/src/view/terminal_render.rs: line_margin_row_text returns an empty string when the label is None; build_line_margin now consults the timeline regardless of the timestamp toggle (it previously only queried it when timestamps were enabled, so number-only mode could not detect blank lines at all).

The margin column width is unchanged, so lines that do have output keep exactly the same alignment as before.

Screenshot

Before	After
Captured in #244 ([ ] on every empty line)	Not attached — needs a quick look in the running app, see below

How to Test

Automated:

cargo test -p terminal --lib line_timeline          # 8 passed
cargo test -p terminal_view --lib view::terminal_render  # 7 passed
cargo test -p terminal --lib                        # 507 passed
cargo test -p terminal_view --lib                   # 566 passed

The behavioural test is margin_blanks_rows_below_the_last_output: it drives a real Term (writes "$ one\r\ntwo\r\n"), then asserts every row below the last output is empty — for the timestamp mode and for the number-only mode. Mutation check: restoring a “render an empty label anyway” path turns that test red (["[] 4 ", "[] 5 ", "[] 6 "]), so the test pins the wiring and not just the pure helper.

Manual (not done by me — please confirm visually):

1. Open an SSH connection with row timestamps and/or line numbers enabled.
2. The area below the cursor stays completely clean, including the line-number column.
3. Lines that have output still show [HH:MM:SS] and/or the line number, aligned exactly as before.

Checklist

• I have read the CONTRIBUTING document and followed the guidelines.
• Reviewed the changes in this PR and confirmed AI generated code (If any) is accurate.
• Passed cargo run for story tests related to the changes.
• Tested macOS, Windows and Linux platforms performance (if the change is platform-specific)