execution, engine: misc fuzzing fixes

[MichaHoffmann]

• check for duplicate labels at every step
• binary operations bail out early if high card side is empty

[MichaHoffmann]

If its too many changes at once we can split the PR too!

[yeya24]

Can we check bench results

[MichaHoffmann]

Can we check bench results

Can do in a bit, but ultimately we need to check for duplicates at every step since thats what prometheus is doing too roughly.

[MichaHoffmann]

LGTM overall, I just wonder if we need to wrap the user defined operator too.

I think generally yes since we have no preconditions to prove otherwise!

[MichaHoffmann]

No meaningful regressiosn in benchmarks:

$ benchstat benchmarks/new.out benchmarks/pr.out
goos: linux
goarch: amd64
pkg: github.com/thanos-io/promql-engine/engine
cpu: 11th Gen Intel(R) Core(TM) i5-1135G7 @ 2.40GHz
                                                     │ benchmarks/new.out │         benchmarks/pr.out          │
                                                     │       sec/op       │    sec/op     vs base              │
RangeQuery/vector_selector-8                                 66.25m ± ∞ ¹   66.10m ± ∞ ¹       ~ (p=1.000 n=5)
RangeQuery/sum-8                                             51.74m ± ∞ ¹   52.32m ± ∞ ¹  +1.12% (p=0.008 n=5)
RangeQuery/sum_by_pod-8                                      60.00m ± ∞ ¹   60.48m ± ∞ ¹       ~ (p=0.310 n=5)
RangeQuery/topk-8                                            50.64m ± ∞ ¹   52.33m ± ∞ ¹  +3.34% (p=0.032 n=5)
RangeQuery/bottomk-8                                         50.23m ± ∞ ¹   52.47m ± ∞ ¹  +4.46% (p=0.008 n=5)
RangeQuery/rate-8                                            112.8m ± ∞ ¹   107.0m ± ∞ ¹       ~ (p=0.151 n=5)
RangeQuery/subquery-8                                        184.6m ± ∞ ¹   184.9m ± ∞ ¹       ~ (p=1.000 n=5)
RangeQuery/sum_rate-8                                        89.82m ± ∞ ¹   93.80m ± ∞ ¹       ~ (p=0.690 n=5)
RangeQuery/sum_by_rate-8                                     107.0m ± ∞ ¹   106.8m ± ∞ ¹       ~ (p=0.690 n=5)
RangeQuery/quantile_with_variable_parameter-8                129.3m ± ∞ ¹   132.6m ± ∞ ¹  +2.56% (p=0.008 n=5)
RangeQuery/binary_operation_with_one_to_one-8                47.83m ± ∞ ¹   48.19m ± ∞ ¹       ~ (p=0.222 n=5)
RangeQuery/binary_operation_with_many_to_one-8               119.2m ± ∞ ¹   119.5m ± ∞ ¹       ~ (p=0.841 n=5)
RangeQuery/binary_operation_with_vector_and_scalar-8        100.24m ± ∞ ¹   99.11m ± ∞ ¹       ~ (p=0.310 n=5)
RangeQuery/unary_negation-8                                  68.91m ± ∞ ¹   69.75m ± ∞ ¹       ~ (p=0.222 n=5)
RangeQuery/vector_and_scalar_comparison-8                    99.44m ± ∞ ¹   99.22m ± ∞ ¹       ~ (p=1.000 n=5)
RangeQuery/positive_offset_vector-8                          64.43m ± ∞ ¹   64.06m ± ∞ ¹       ~ (p=0.841 n=5)
RangeQuery/at_modifier_-8                                    58.44m ± ∞ ¹   57.52m ± ∞ ¹       ~ (p=0.310 n=5)
RangeQuery/at_modifier_with_positive_offset_vector-8         56.09m ± ∞ ¹   55.94m ± ∞ ¹       ~ (p=0.690 n=5)
RangeQuery/clamp-8                                           85.71m ± ∞ ¹   87.20m ± ∞ ¹       ~ (p=0.095 n=5)
RangeQuery/clamp_min-8                                       80.25m ± ∞ ¹   80.98m ± ∞ ¹  +0.92% (p=0.016 n=5)
RangeQuery/complex_func_query-8                              116.4m ± ∞ ¹   126.0m ± ∞ ¹  +8.26% (p=0.008 n=5)
RangeQuery/func_within_func_query-8                          111.9m ± ∞ ¹   110.3m ± ∞ ¹       ~ (p=0.310 n=5)
RangeQuery/aggr_within_func_query-8                          117.5m ± ∞ ¹   116.4m ± ∞ ¹       ~ (p=0.548 n=5)
RangeQuery/histogram_quantile-8                              338.8m ± ∞ ¹   340.0m ± ∞ ¹       ~ (p=1.000 n=5)
RangeQuery/sort-8                                            68.22m ± ∞ ¹   67.38m ± ∞ ¹       ~ (p=0.095 n=5)
RangeQuery/sort_desc-8                                       67.71m ± ∞ ¹   67.65m ± ∞ ¹       ~ (p=0.690 n=5)
RangeQuery/absent_and_exists-8                               47.78m ± ∞ ¹   48.55m ± ∞ ¹       ~ (p=0.056 n=5)
RangeQuery/absent_and_doesnt_exist-8                         1.113m ± ∞ ¹   1.129m ± ∞ ¹  +1.45% (p=0.032 n=5)
NativeHistograms/selector-8                                  427.6m ± ∞ ¹   426.1m ± ∞ ¹       ~ (p=0.690 n=5)
NativeHistograms/sum-8                                       502.1m ± ∞ ¹   510.3m ± ∞ ¹  +1.64% (p=0.032 n=5)
NativeHistograms/rate-8                                      923.8m ± ∞ ¹   979.8m ± ∞ ¹       ~ (p=0.548 n=5)
NativeHistograms/sum_rate-8                                   1.022 ± ∞ ¹    1.013 ± ∞ ¹       ~ (p=0.548 n=5)
NativeHistograms/histogram_sum-8                             866.1m ± ∞ ¹   877.0m ± ∞ ¹       ~ (p=0.222 n=5)
NativeHistograms/histogram_count-8                           862.5m ± ∞ ¹   870.8m ± ∞ ¹       ~ (p=0.548 n=5)
NativeHistograms/histogram_quantile-8                        553.2m ± ∞ ¹   543.3m ± ∞ ¹       ~ (p=0.690 n=5)
NativeHistograms/histogram_scalar_binop-8                    906.5m ± ∞ ¹   903.8m ± ∞ ¹       ~ (p=0.548 n=5)
geomean                                                      120.2m         121.0m        +0.72%
¹ need >= 6 samples for confidence interval at level 0.95