feat: report per-ingredient page occupancy

[MichaReiser]

I tried to get more insight into #1179, specifically if there are large differences by ingredient. For this, I added per-ingredient page occupancy through Salsa's unstable memory-usage API, including non-empty page count, capacity, excess capacity, and P25/P50/P75/P90/P99 fill counts.

Empty pages are excluded, because they don't materialize (they're MaybeInit slices) until the first item is written to the page.

[netlify]

👷 Deploy Preview for salsa-rs processing.

Name	Link
🔨 Latest commit	4fb0eb8
🔍 Latest deploy log	https://app.netlify.com/projects/salsa-rs/deploys/6a3a5aee419396000861a82e

[netlify]

✅ Deploy Preview for salsa-rs canceled.

Name	Link
🔨 Latest commit	55b0361
🔍 Latest deploy log	https://app.netlify.com/projects/salsa-rs/deploys/6a3b85ed2c2e67000823dff8

[codspeed-hq]

Merging this PR will degrade performance by 7.48%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 4 (👁 1) regressed benchmarks
✅ 38 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	amortized[InternedInput]	1.8 µs	2 µs	-8.76%
❌	Simulation	cached[Input]	1.8 µs	1.9 µs	-7.52%
❌	Simulation	amortized[Input]	1.9 µs	2 µs	-7.19%
👁	Simulation	amortized[SupertypeInput]	2.5 µs	2.7 µs	-6.43%

Tip

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

---

_{Comparing MichaReiser:page-occupancy-metrics (55b0361) with master (e769fa0)}