TypeScript indexer doesn't emit the language field scip.language

[donsbot]

When combining SCIP data from many sources it is very useful to know the language of the file.

message Document {
  string language = 4;

I believe scip-typescript may be setting this to "" .

there's also something funky with non-utf8 in the symbol fields.

$ protoc --decode scip.Occurrence scip.proto < /tmp/index.scip
E0626 17:07:01.064803 1569665 wire_format_lite.cc:626] String field 'scip.Occurrence.symbol' contains invalid UTF-8 data when parsing a protocol buffer. Use the 'bytes' type if you intend to send raw bytes.