[Releases 2.13] Output hybrid error message in marqo logs

[vicilliar]

• What kind of change does this PR introduce? (Bug fix, feature, docs update, ...)
  bug fix

• What is the current behavior? (You can also link to an open issue here)
  any errors in tensor or lexical search in hybrid rrf search are swallowed by java code

• What is the new behavior (if this is a feature change)?
  output any tensor or lexical errors in marqo logs

• Does this PR introduce a breaking change? (What changes might users need to make in their application due to this PR?)
  no

• Have unit tests been run against this PR? (Has there also been any additional testing?)

• Related Python client changes (link commit/PR here)

• Related documentation changes (link commit/PR here)

• Other information:

• Please check if the PR fulfills these requirements

• The commit message follows our guidelines
• Tests for the changes have been added (for bug fixes/features)
• Docs have been added / updated (for bug fixes / features)

[papa99do]

Please check with Vespa team about the best practices of handling errors in the search chain.

[papa99do]

I was wondering if we should handle errors differently after reading Vespa docs. Instead of raising Runtime exception everywhere (which cause Vespa to always return 500). We should probably leverage HitGroup().addError method which is suggested in this doc.

For this specific scenario, we are collecting errors from both lexical result and tensor result. So we can do this:

HitGroup combinedErrors = new HitGroup();
combinedErrors.addErrorsFrom(resultTensor.hits());
combinedErrors.addErrorsFrom(resultLexical.hits());
if (combinedErrors.getError() != null) {
      return new Result(query, combinedErrors);
}

In this way, you should be able to collect all the errors from both result and allow search chain to handle the error properly, instead of just concatenating the first error found in each result.