ci: PyPy workers on the ALL tier grow until they are OOM-killed

On PyPy, the unit ALL tier's pytest workers grow to roughly 3.5–4.5 GB each.

Symptom: in a 14 GB container at -n 4, four workers were OOM-killed, at different tests on each run. Capping the heap at 3 GB only turned the kills into MemoryErrors.
Expected on CI: a 16 GB GitHub runner is likely to hit the same limit, so the PyPy legs added for ci: categorise the tests into tiers and run each tier where it belongs #1958 phase 3 need containment before they can block.
Found on PyPy 8.0.1, run in an Ubuntu 24.04 container.
Possible fixes:
- split the tier into several pytest processes, so each worker starts fresh;
- recycle workers after a fixed number of tests;
- lower -n on PyPy.
Whichever is chosen, measure peak RSS per worker before and after.

ci: run the curated ALL tier on PyPy, split into separate pytest runs (…

The fix is folded into draft #2006 (d4db362e0). The growth is steady across tests, not a handful of heavy modules, so lowering -n or setting PYPY_GC_MAX cannot help. A PyPy leg instead runs the tier as separate pytest invocations, each with fresh workers. In a 14 GB container, workers now peak at 2.6 GB and the container at 8.5 GB, with no OOM kills.