chore: add query and reflection benchmarks

[makspll]

Summary

• Adds more interesting benchmarks around querying and reflection, revealing slower parts of BMS and/or Bevy
• Re-create all plots based on benchmarks present every push to main

[Copilot]

Pull Request Overview

This PR adds new benchmarks for query and reflection to highlight slower aspects of BMS and/or Bevy. Key changes include updating the Xtasks::Bench variant to accept a publish flag, adjusting the corresponding bench function signature and behavior, and adding pre-benchmark hooks and random function registrations in the test harness.

Reviewed Changes

Copilot reviewed 4 out of 20 changed files in this pull request and generated no comments.

File	Description
crates/xtask/src/main.rs	Updated Xtasks::Bench variant and bench function to handle a publish flag
crates/testing_crates/script_integration_test_harness/src/lib.rs	Added pre_bench hook calls in Lua and Rhai benchmark routines
crates/testing_crates/script_integration_test_harness/src/test_functions.rs	Added random and reseed functions for test utilities
crates/testing_crates/script_integration_test_harness/Cargo.toml	Added rand and rand_chacha dependencies to support new functions

Files not reviewed (16)

• assets/benchmarks/function/call.lua: Language not supported
• assets/benchmarks/function/call.rhai: Language not supported
• assets/benchmarks/function/call_4_args.lua: Language not supported
• assets/benchmarks/function/call_4_args.rhai: Language not supported
• assets/benchmarks/math/vec_mat_ops.lua: Language not supported
• assets/benchmarks/math/vec_mat_ops.rhai: Language not supported
• assets/benchmarks/query/1000_entities.lua: Language not supported
• assets/benchmarks/query/1000_entities.rhai: Language not supported
• assets/benchmarks/query/100_entities.lua: Language not supported
• assets/benchmarks/query/100_entities.rhai: Language not supported
• assets/benchmarks/query/10_entities.lua: Language not supported
• assets/benchmarks/query/10_entities.rhai: Language not supported
• assets/benchmarks/reflection/10.lua: Language not supported
• assets/benchmarks/reflection/10.rhai: Language not supported
• assets/benchmarks/reflection/100.lua: Language not supported
• assets/benchmarks/reflection/100.rhai: Language not supported

Comments suppressed due to low confidence (3)

crates/xtask/src/main.rs:726

• [nitpick] The use of the name 'execute' for the bench parameter here is inconsistent with the 'publish' field in the enum. Consider renaming 'execute' to 'publish' for clarity and consistency.

Xtasks::Bench { publish: execute } => Self::bench(app_settings, execute),

crates/xtask/src/main.rs:1226

• [nitpick] The function parameter 'execute' could be renamed to 'publish' to match the enum variant and clearly convey its purpose in controlling benchmark publishing versus dry-run behavior.

fn bench(app_settings: GlobalArgs, execute: bool) -> Result<()> {

crates/testing_crates/script_integration_test_harness/src/lib.rs:358

• The pre_bencher function is retrieved with .ok() and then immediately unwrapped later. Consider handling potential errors from the pre_bencher call to avoid unexpected panics during benchmarking.

let pre_bencher: Option<Function> = ctxt.globals().get("pre_bench").ok();