Add an e2e perf/load test with blocking commits

[eddyashton]

Actual throughput isn't so interesting right now, although it does look satisfyingly stable.

Instead this should be seen as an initial load/stress test, confirming that the nodes and task system can handle many concurrent users without catastrophe/segfault.

Critically it requires many parallel clients, each waiting for responses rather than pipelining, to get sensible results, hence the recent improvements to basicperf.py.

[Copilot]

Pull request overview

Adds a new end-to-end load/perf scenario which blocks responses until consensus commit, intended as a stress test for concurrency/task handling rather than raw throughput.

Changes:

• Extend basicperf.py with blocking_write/blocking_read workloads targeting new “blocking” record endpoints.
• Improve basicperf.py stacked-bar visualisation by binning many clients into visible groups and adjusting color gradients.
• Add blocking record endpoints to the basic sample app and register a new pi_basic_blocking perf test in CMake.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
tests/infra/basicperf.py	Adds blocking workload generators and improves plotting for many concurrent clients.
samples/apps/basic/basic.cpp	Adds /records/blocking/{key} GET/PUT endpoints which respond only once the tx is committed.
CMakeLists.txt	Registers a new perf configuration test exercising many concurrent blocking writers.