semanticTokens/full timeout too fast, need an option to increase latency to be beyond 5 seconds

(tested on Scala 2.13 + metaLS)

this is part of the log on my project:

[Trace - 20:21:19] Sending request 'textDocument/semanticTokens/full - (46)'.
Params: {
  "textDocument": {
    "uri": "file:///home/peng/git/spookystuff/prover-commons/module/core/src/main/scala/ai/acyclic/prover/commons/graph/Engine.scala"
  }
}


[Trace - 20:21:24] Sending notification '$/cancelRequest'
Params: {
  "id": "46"
}


[Trace - 20:21:24] Received response 'textDocument/semanticTokens/full - (46)' in 4935ms.
No response returned.
Error: {
  "code": -32800,
  "message": "The request (id: 46, method: \u0027textDocument/semanticTokens/full\u0027) has been cancelled"
}


[Trace - 20:21:24] Sending request 'textDocument/semanticTokens/full - (47)'.
Params: {
  "textDocument": {
    "uri": "file:///home/peng/git/spookystuff/prover-commons/module/core/src/main/scala/ai/acyclic/prover/commons/graph/Engine.scala"
  }
}


[Trace - 20:21:57] Sending notification '$/cancelRequest'
Params: {
  "id": "47"
}


[Trace - 20:21:57] Received response 'textDocument/semanticTokens/full - (47)' in 32945ms.
No response returned.
Error: {
  "code": -32800,
  "message": "The request (id: 47, method: \u0027textDocument/semanticTokens/full\u0027) has been cancelled"
}


[Trace - 20:21:57] Sending request 'textDocument/semanticTokens/full - (48)'.
Params: {
  "textDocument": {
    "uri": "file:///home/peng/git/spookystuff/prover-commons/module/core/src/main/scala/ai/acyclic/prover/commons/graph/Engine.scala"
  }
}

after 5 seconds the request will be canceled without a reason and replaced with a new request.

Some languages compiles slower than others, we probably need a new workflow to ensure that ongoing request won't timeout prematurely

Start by tracing the LSP client's handling of textDocument/semanticTokens/full and the $/cancelRequest notifications shown in the report. Determine where the five-second cancellation is enforced and how an increased or configurable latency should preserve ongoing requests; done means slow semantic-token requests are not canceled prematurely and the behavior is covered by the project's relevant checks.