use sqlCtx in handler

[jycor]

Tests here: dolthub/dolt#8493

[zachmu]

This looks right to me but it could use some more tests, very easy to get this stuff wrong. I see the test in Dolt but it would be nice to have one a Go test in GMS as well

[reltuk]

This function is a mess. I agree that the same context that gets returned from BeginQuery should be used as the argument to ProcessList.EndQuery here and as the context argument to queryExec. But, I don't agree with a lot of details of this current function, including the code directly touched here:

1. It seems like the err on line 422 should be checked.
2. It seems like EndQuery should be called regardless of whether there was an error.
3. Checking for ctx != nil in the defer block seems wrong.
4. On line 414 we defer finalize(err), which is immediately evaluating the err variable, where as it seems like we meant to defer the evaluation until we know the error return of the actual function.
5. Given that we want to defer evaluating the return value of this function it seems like we shouldn't shadow err on line 428. And, because we use a direct return of the callback evaluation on line 479, we either need to change that to err = callback... or we need to switch to named return values.

Made these cleanups here if you agree:

#2723

Otherwise looks good to me. Definitely an appropriate path to fixing the kill behavior we've seen. Good findings and investigation.