Block responding to textDocument/semanticTokens until schema is available

[radeksimko]

Current Version

0.11.0

Use-cases

As mentioned in #331:

Client always sends textDocument/semanticTokens/full almost immediately after didOpen and didChange, which means that we may not have all the schema at that point. As a result not supporting this call initially may provide occasionally "flickering UX" (colours change as the user types), but the upside is that it can't be any worse than it is, because we are just enriching the highlighting and the static grammar in VSCode extension still provides the base.

Proposal

Another way of resolving this would be also pushing update from the server back to client when we have the schema via workspace/semanticTokens/refresh, but it's questionable whether this is the best use case for that method.

I believe that textDocument/semanticTokens should always attempt to wait for some sensible time until schema is available, or timeout and return no (additional) tokens. This would also avoid the extra roundtrip in most cases.

[radeksimko]

#344 may be better way to resolve this than effectively delay rendering / response, but I'll keep this open until we have at least prototype of #344 to fully understand the UX implications.