Add support for git-crypt

[mikeseese]

This PR resolves #34330 by adding support for git-crypt, a tool which encrypts and decrypts files as a git filter.

Currently VS Code does not properly show diffs for files that are encrypted on the remote but decrypted locally.

Here is some of the current behavior:

A file will have no changes (see below that production.yaml doesn't show up in the git status panel) but will show in the editor that the whole file was modified:

We can see here when bringing up the inline diffs, that it's comparing the decrypted local file with the encrypted HEAD version:

Lastly, when the file is changed (shown below as showing up in the git status on the panel), and the git diff is opened, the diff isn't shown because the left side is the binary version:

This PR is fairly simple, and doesn't attempt to solve the total scope of adding generic support for any and all filters. However, I hope this can get accepted in the meantime before that solution is realized to support this specific issue.

This PR checks after a git show to see if the buffer contains a magic word that indicates it is a git-crypt encrypted blob. If it is, then it runs it into git-crypt smudge which will decrypt the contents appropriately. This resolves all of the cases mentioned above.

A potential implementation for solving the generic issue would be to read .gitattributes for the file being git show'd and see what the filter is and hand that as the program to execute the blob on. However, this would require significantly more testing which was out of scope for the work I'm doing.

[martin-css]

This looks like a git-crypt specific version of #81395 which I'd coincidentally submitted a few day prior.

This approach is certainly simpler whereas mine was focused more on support for a generic git filter. Regardless of whatever is thought to be best, would be nice to have git-crypt file diffs working properly!

[mikeseese]

I would definitely prefer a generic solution rather than a specific one. Nice find (of this PR) @martin-css! I'll take a look at your PR

[mikeseese]

Closing in favor of #81395