Proposal: support Deno

[sigmaSd]

Let me get started with its just a proposal because I like Deno and obviously you can just ignore this.

Ok whats it this:

• This PR makes the current codebase works with deno, there are 2 main commits here
• • the first one https://github.com/leona/helix-gpt/commit/a237f6c62305c77c619b733435745d098dde5775 makes the codebase works with deno by injecting a Bun global, but it only works if deno is run like this deno run --unstable-sloppy-imports --unstable-bare-node-builtins src/app.ts , this works, but it have 2 issues, the deno lsp will keep giving errors, and it can't be compiled curretnly (there is an active issue in deno)
• • the second commit https://github.com/leona/helix-gpt/commit/51dc3a9883b03d77ba4b73492e97eb3a03912d3a converts bare specifiers to specifiers that ends with .ts this works in both deno and bun, and it have the advantage of fixing the 2 above issues, you can now just run it with deno run and it can be compiled with deno

Maybe some advantages:

• https://jsr.io/ have made me think thats its nice to support multiple runtimes, and I think its the case here
• you get to use deno cool permisison model, for example the current code base only require --allow-env --allow-net thats a secure binary (no read or write)
• Deno wins this round for compile size the compiled app size is 74mb vs bun smol 90mb and there is a proposal to make the user compile its own specific binary which would make this even smaller https://github.com/denoland/deno/issues/22824

[sigmaSd]

Actually maybe its better to keep this as a fork to not add unneeded complexity to this project, unless you think otherwise