[Repo Assist] perf: avoid Seq allocations in hot markdown-parsing paths

[github-actions]

🤖 This is an automated pull request from Repo Assist.

Summary

Three small but impactful performance improvements in parsing hot paths, eliminating unnecessary Seq enumerator allocations and object boxing.

1. removeSpaces — count leading whitespace without Seq

File: src/Common/StringParsing.fs

removeSpaces is called for every XML doc comment and every literate code block. The old implementation counted leading whitespace with:

line |> Seq.takeWhile Char.IsWhiteSpace |> Seq.length

This allocates two IEnumerator objects per non-empty line and boxes each char. The replacement uses line.Length - line.TrimStart().Length, which is a single O(n) pass with no allocations.

2. StartsWithNTimesTrimIgnoreStartWhitespace — count fence chars without Seq.windowed

File: src/Common/StringParsing.fs

This active pattern is called for every line during markdown block parsing to detect fenced code blocks (``` or ~~~). The old implementation:

Seq.windowed start.Length startAndRest
|> Seq.map (fun chars -> System.String(chars))
|> Seq.takeWhile ((=) start)
|> Seq.length

...built a sliding-window sequence over the entire line, allocated a new System.String for each window, and counted with Seq.length. For a 100-character line with a 3-char fence, that's ~98 string allocations before yielding 3.

Replaced with a direct index loop:

let mutable count = 0
let mutable pos = 0
let startLen = start.Length
while pos + startLen <= startAndRest.Length
      && startAndRest.Substring(pos, startLen) = start do
    count <- count + 1
    pos <- pos + startLen
count

3. readXmlElementAsSingleSummary — count leading spaces without Seq

File: src/FSharp.Formatting.ApiDocs/XmlDocReader.fs

Same Seq.takeWhile + Seq.length pattern, called once per line per XML doc comment to detect indentation uniformity. Replaced with line.Length - line.TrimStart([|' '|]).Length.

Test Status

• ✅ dotnet build FSharp.Formatting.sln --configuration Release — 0 warnings, 0 errors
• ✅ 317 Markdown tests pass
• ✅ 143 Literate tests pass
• ✅ 30 CodeFormat tests pass
• ✅ 29 ApiDocs tests pass
• ✅ All 520 tests pass — no regressions

Generated by 🌈 Repo Assist, see workflow run. Learn more.

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@97143ac59cb3a13ef2a77581f929f06719c7402a

Generated by 🌈 Repo Assist, see workflow run. Learn more.

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@97143ac59cb3a13ef2a77581f929f06719c7402a