Sorry about this! However, this diff reintroduces the warning described in ENG-11162. Instead, I just updated the import in index.js: b1ca7d1

Ready for review

Overall looks good, but I have a lot of comments so requesting changes

Unfortunately I had to introduce a lot of boilerplate due to changes in Emscripten, but the CI builds all pass now. I have some more work to do before this is ready... see inline comments

Avoid @generated in run_emscripten.sh in code that adds it to comm-query-executor.js

Simplify @generated sed command

How do we handle converting in the other direction? Ie. when a user uses the reply function in Comm for a Farcaster thread, we send the reply info correctly to the Farcaster backend. This part seems like it might be a bit harder since I think the Farcaster backend tracks replies based on message IDs instead of just simple Markdown

Emscripten and Farcaster signer CI workflows are expected to fail for the same reason as in the parent diff.

Buildkite CI jobs that rely on Nix (Emscripten, Farcaster signer) are expected to fail, since we run the Buildkite agent inside of Nix. After landing this I'll need to kill the Buildkite agent, exit the Nix environment, update the repo, and then re-run nix develop and the Buildkite agent. Improving this is tracked in ENG-6439.

Here are some edits to the copy. Let me know if any of it is too long to fit in the UI!

What would it take to add Chrome web support?