Query/Perf: don't compile liftable constant resolvers in interpretation mode when the resolver itself contains a lambda #35208
Description
This is part of a bigger perf regression between EF8 and EF9: #35053
When processing LiftableConstantExpressions in the regular (non-AOT) mode we compile the resolver lambda and then evaluate it to get back the actual constant we plan to use. We do the compilation in the interpretation mode and that causes problems if the resulting object is itself (or contains) a delegate. Compiling into a delegate using interpreted mode, has impact on allocations as well as execution speed.
before the fix (this already includes the invoke fix #35206)
| Method | Async | Mean | Error | StdDev | Op/s | Gen0 | Gen1 | Allocated |
|---|---|---|---|---|---|---|---|---|
| MultiInclue | False | 487.1 ms | 0.99 ms | 0.88 ms | 2.053 | 17000.0000 | 11000.0000 | 103.29 MB |
| MultiInclue | True | 487.4 ms | 3.63 ms | 3.22 ms | 2.052 | 17000.0000 | 11000.0000 | 103.29 MB |
after the fix
| Method | Async | Mean | Error | StdDev | Op/s | Gen0 | Gen1 | Allocated |
|---|---|---|---|---|---|---|---|---|
| MultiInclue | False | 455.1 ms | 8.94 ms | 10.29 ms | 2.197 | 11000.0000 | 6000.0000 | 67.92 MB |
| MultiInclue | True | 435.4 ms | 1.77 ms | 1.66 ms | 2.297 | 11000.0000 | 6000.0000 | 67.92 MB |