Base Test Classes are inefficient

parseq-test has a base test class that creates an engine and threadpool per test class instead of one shared one overall. The former is excellent for testing parseq itself in isolation, but the latter is better for testing usages of parseq in that it doesn't create hundreds of thousands of threads.

Happy to contribute a patch too.

Start by locating the base test classes in parseq-test and inspect how the engine and thread pool are created for each test class. The change is complete when suitable usage tests share one engine and thread pool instead of creating them repeatedly, while parseq's own isolation tests remain supported.

Written by the indexing model from the issue text.