[callout]: preserve spaces in named titles

[evnluo]

Fixes #605

Summary

• recover named callout values from the joined Hexo tag arguments after Hexo has stripped quote characters
• preserve spaces in title values while keeping named type, icon, and class parsing intact

Verification

• The repository has no test runner; the PR build workflow will run npm install and npm run build.
• No generated build artifacts are included.

[github-actions]

Preview: https://redefine-preview-pr-618.vercel.app

[evannotfound]

@evnluo Any potential code simplification to be made and make it cleaner?

[evnluo]

I reviewed the change. The recovery approach is reasonable, but I’d make two small cleanups before merging:

1. Parse the raw named-argument positions once. getCalloutNamedValue() currently calls findNamedArgs(rawArgs) separately for each field. Compute const rawNamedArgs = findNamedArgs(rawArgs) once in parseNamedArgs() and pass it into getRawNamedValue(). That keeps the behavior the same while removing repeated scans and making the data flow easier to follow.
2. Use CALLOUT_NAMED_KEYS as the single source of truth for both the boundary detection and supportsNamed. The six keys are currently listed twice.

I’d keep the quote-recovery state machine local to callout.js. It is clearer than trying to make the shared tag parser infer boundaries after Hexo has already split the arguments, and I wouldn’t add a broader abstraction for this one caller. The state machine itself is worth keeping because a regex-only solution would be fragile around quoted or escaped values.

So I don’t see a behavior change that’s needed, just those two refactors to make the added code easier to maintain. Since the repo has no test runner, I’d manually check spaced and quoted title values alongside named type, icon, and class, then use npm run build for validation.

[evannotfound]

@evnluo review and implement the potential simplification