Get benchmark node memory stats from /proc

[eddyashton]

Follow-up to #7822. We no longer need a built-in /node/memory endpoint, since these stats can be gathered for current platforms from standard tools. This uses those standard tools (reading from /proc/{pid}/status) to get memory use during our e2e benchmark runs, restoring those values going to Bencher.

[Copilot]

Pull request overview

Replaces the removed /node/memory endpoint by collecting node memory metrics directly from /proc/<pid>/status during e2e benchmark runs (and including per-test memory snapshots in the e2e suite report), so these values can continue to be published to Bencher on supported platforms.

Changes:

• Add /proc/<pid>/status parsing helper to extract RSS/peak RSS/virtual size in bytes.
• Publish primary-node memory metrics to Bencher in benchmark runners.
• Record per-node memory stats in e2e_suite.py test run reporting.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
tests/infra/proc.py	Adds get_proc_memory_stats() to read RSS/HWM/VmSize from /proc.
tests/infra/runner.py	Collects primary PID memory stats and publishes them to Bencher after benchmark runs.
tests/infra/piccolo_driver.py	Collects primary PID memory stats and publishes them to Bencher after benchmark runs.
tests/infra/basicperf.py	Collects primary PID memory stats and publishes them to Bencher (when primary wasn’t stopped).
tests/e2e_suite.py	Adds mem_stats() and includes memory info in per-test JSON reporting.