[NFR]: Benchmark subjects for components with no reference app #17695

Summary

The benchmark harness in tests/benchmarks/ measures the 4 reference apps (MVC, Micro, REST, ADR) and the components that they use: Di, Filter, Http, Mvc, Db, Events, Support, Container and ADR. The other components have no subjects and no measurements.

This issue adds subjects for 9 of these components. Then the performance work can rank and change them the same way as the components of the apps.

Scope

23 new subjects in 10 classes (11 files, tests/benchmarks/ only):

• The subjects use no network: Session uses the Noop adapter, Cache the memory adapter, and Logger a small bench adapter with no I/O (Logger\LastLineAdapter).
• Each subject checks its result and throws an exception if the result is wrong.
• An A/A run on the same build must show no change, so that each subject does the same work on each call.
• The README lists the new subjects and the run-all batches.

Not in scope

• No change to the framework code (phalcon/, ext/).
• Optimizations of these components: each one gets its own issue after the measurements.
• Paginator, Annotations, Translate, Assets, Html\TagFactory, Tag, Image, and the Redis and Memcached adapters (a later batch).