Bind expressions before optimization

[mhk197]

Instead of optimizing expressions and then binding, first bind and then optimize.

There are many benefits to this, including:

• We only optimize well-typed expression trees (we dont waste work on invalid trees that fail at execution time). Rewrites cannot obscure typing errors.
• We no longer have multiple return_dtype walks that recurse through the entire expression tree, and we dont need an optimizer-owned dtype cache (SimplifyCache).
• Few APIs and duplicate rules
• Preparation for lambdas and variables. Variables must be typed before optimization. The optimizer can treat lambda scopes as opaque.
• We can recursively optimize ONCE rather than incrementally calling expression.optimize on construction of every tree node.

[codspeed-hq]

Merging this PR will degrade performance by 7.04%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚡ 2 improved benchmarks
❌ 7 regressed benchmarks
✅ 1972 untouched benchmarks
⏩ 442 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	optimize_lookup_predicate[ids=1, shape=balanced_or]	47 µs	59.4 µs	-20.91%
❌	Simulation	optimize_lookup_predicate[ids=64, shape=in_list]	45.2 µs	56.1 µs	-19.5%
❌	Simulation	optimize_lookup_predicate[ids=128, shape=in_list]	45.4 µs	56.3 µs	-19.44%
❌	Simulation	optimize_lookup_predicate[ids=256, shape=in_list]	46.5 µs	57.1 µs	-18.57%
❌	Simulation	optimize_lookup_predicate[ids=192, shape=in_list]	46.2 µs	55.9 µs	-17.36%
❌	Simulation	optimize_lookup_predicate[ids=16, shape=in_list]	45.5 µs	54.6 µs	-16.74%
❌	Simulation	optimize_lookup_predicate[ids=16, shape=balanced_or]	88.1 µs	102.9 µs	-14.35%
⚡	Simulation	optimize_lookup_predicate[ids=1, shape=in_list]	151.3 µs	80.3 µs	+88.42%
⚡	Simulation	compress_fsst[(500, 64, 8)]	585.9 µs	524.3 µs	+11.75%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing mk/bound-expression-optimize (c296e37) with develop (46a8d39)}

Footnotes

1. 442 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩