I'd prefer not to add all those extra crates if we can avoid it. I'll follow up on the initial issue.

I've reverted the crate addition/usage, with another suggested method: trimming potential delimiters from non-freeform metadata lines

I wonder if we could perhaps hackily solve this by stripping off trailing commas from comment lines, as a concession to spreadsheet users?

Instead of just removing commas, it makes use of the fact that the currently supported delimiters are all non-ascii-alphanumeric. This would allow samples like #separator:,,,, #separator:Pipe|||, #html:true,,, and #tags column:8,,, to be parsed instead of being ignored, as demonstrated in the test

All the freeform metadata values (#deck, #notetype, #tags, #columns) are left alone to avoid breakage, at the expense of inconsistency with the other lines. #match scope and #if match are left alone because their constrained values can include or +, which unfortunately aren't compatible with this method

One problem with such selective/"magical" approaches is that they can cause confusion. A user may find a simple metadata line in a spreadsheet works, then wonder why other ones don't work correctly when they go to use them in the future. But indiscriminately stripping non-alphanumeric characters from all metadata lines is going to confuse/inconvenience some users too, so it's not an easy call.

Given that your approach is a subset of the other one and doesn't preclude us doing more later if we feel we need to, it seems like the more conservative choice. Thank you @iamllama!