Adds more benchmarks

[ncordon]

Why

To test out more improvements on auto-completion and signature help, especially when we are moving the cursor in the query

[ncordon]

This doesn't have much effect in the results of the benchmark, except on the parsing time, which I believe is cause the first iteration is always going to be really slow

[OskarDamkjaer]

I could be wrong but I think the semantic analysis for the really large query takes more than a second per cycle, so won't setting the maxTime effectively make that benchmark useless?

[ncordon]

With the 1s limit:

simple - validate syntax x 5,016 ops/sec ±4.13% (26 runs sampled)
movies - validate syntax x 0.38 ops/sec ±2.84% (10 runs sampled)
tictactoe - validate syntax x 126 ops/sec ±6.65% (23 runs sampled)

Without:

simple - validate syntax x 4,893 ops/sec ±1.38% (93 runs sampled)
movies - validate syntax x 0.39 ops/sec ±2.01% (10 runs sampled)
tictactoe - validate syntax x 124 ops/sec ±3.25% (74 runs sampled)

I don't think that 1s is a hard limit, so if the benchmarks overrun that it detects it well