Concatenate the text segments of image alt text

[bobzhang]

Split out of #152. Found by differential fuzzing against markdown-it-py (commonmark preset).

Bug: the HTML renderer joined the plain-text segments of each line of an image description with newlines, and joined the lines with nothing. That is the reverse of what's intended.

Repro: ![foo *bar*](/u) gave alt="foo \nbar". Expected alt="foo bar".

Fix: concatenate the segments of each line, and separate lines (hard breaks) with newlines, reusing joined_string with Iter::join.

Spec examples fixed: 520, 573, 574, 575, 576, 577, 585, 589.

Tests: src/cmark_html/image_alt_text_test.mbt.

Depends on: nothing, independent of the other PRs.

[bobzhang]

Automated review: Codex CLI (codex exec, model reasoning effort high, read-only sandbox), 1 round, on git diff against the base branch.

Verdict: signed off, no changes requested.

I sign off on the code change. No concrete correctness or regression issues found.

The fix concatenates inline segments and preserves hard-break boundaries, consistent with CommonMark 0.31.2’s image-description guidance. HTML escaping remains intact, and Iter::join is appropriate. The tests cover the reported bug, nested links/images, and hard breaks.

Codex could not run moon test in its read-only sandbox. The tests were run separately on native, js and wasm-gc and pass.