Parallelisation of tracking

Running all recipes in series at high resolution will take too long. So at some point we need to figure out the best way to parallelise (at least the identification), noting that multiple read access to the same file may cause problems.

No files, tests, or entry points are named. Start by locating recipe execution and identification, then determine how concurrent reads of the same file behave; done means an agreed parallelisation approach for at least identification that avoids or handles those read-access problems.

Written by the indexing model from the issue text.