This MR goal is to verify how much works and how much doesn't ... it also already underlines some mistake in coincident, most notably:
the fact the code is the same for both main and worker makes very little sense ... sabayon does a better job as module and it's also easier to reason about and maintain ... that plus the fact workers with embedded coincident would need less code to parse to be to bootstrap
the whole Proxy dance is super ugly and js-proxy would be a much nicer, cleaner / elegant solution + it makes it relatively easy to eventually avoid issues with foreign interpreters such as Pyodide and their way to drill properties without checks
the CHANNEL dance is very elegant and simple in sabayon and, most importantly, is not hard coded ... I believe coincident should actually do pretty much the exact same dance 'cause it's super easy to reason about, less collision prone and, most importantly, it's more secure as it's kept as a secret
the refactoring might, at its initial stage, rule out the server orchestration as that also needs some refactoring to be better, more secure, and not collision prones with multiple pages using multiple, different, servers behind the scene
ditch CommonJS as sabayon needs workers to be modules due its top-level await requirement ... this should be a no-brainer as nobody on the Web should still use CommonJS for anything ... ESM is all we need here
This MR goal is to verify how much works and how much doesn't ... it also already underlines some mistake in coincident, most notably:
server
orchestration as that also needs some refactoring to be better, more secure, and not collision prones with multiple pages using multiple, different, servers behind the scene