Fix sync status and diff handling for branch/subfolder text files

[lukelzlz]

What changed

• preserve the configured branch when fetching remote file contents during sync
• mark sync results as failed when pull operations return file errors
• resolve diff-view remote paths inside configured repository subfolders
• auto-detect common East Asian text encodings when decoding remote GitHub file contents

Why

Sync and diff behavior could be misleading in a few important cases:

• branch-specific files could be fetched from the default branch instead of the configured branch
• sync logs could report completion even when remote pulls failed
• diff view could request the wrong remote path when the vault mapped into a repo subfolder
• non-UTF-8 remote notes could render as mojibake in diff view

Impact

• sync status is now more accurate
• diff view shows the correct remote file in subfolder-based setups
• remote text previews handle more real-world note encodings

Validation

• npm test
• npm run build

[rhoades-brown]

Hi. @lukelzlz - sorry it has taken me so long to review this, I had not seen the GitHub notifications. I am taking a look right now...

[rhoades-brown]

Firstly, thanks again @lukelzlz for the PR, I really appreciate it and some of these bugs are something I came here to work on.

I have cherry-picked the, pull-failure, subfolder-path, and UTF-8 decodeBase64 (instead of atob) fixes have into #25 so they can be applied now.

This PR can keep the GB18030/Big5/Shift-JIS autodetection so we can discuss that separately.

From what I understand, Obsidian only supports UTF-8 - Encoding Auto-Fix is a useful plugin designed to support other formats, but it is not a straight GB18030 loop:

• it converts local files to UTF-8 with BOM/confidence checks
• it does not list GB18030, Big5, or Shift-JIS
• it refuses low-confidence guesses instead of taking the first decoder that does not throw

I understand what you are trying to do, but I wonder if this aspect needs some improvement?

[lukelzlz]

Hi @rhoades-brown,

That makes total sense. Keeping it strictly UTF-8 avoids false-positive decoding and stays aligned with Obsidian's core expectations, so let's skip the multi-encoding try-catch fallback.

I saw that #25 has already been merged—thanks for quickly reviewing and integrating the fixes!

I'll go ahead and close this PR now. Thanks again for maintaining the project!