Misc minor bugs (#142)

* improvements to remove_instructions

* test remove instructions

* correct the function qualifier for Cuda target device functions

* adds missing options in add_prefetch

* function_qualifier->preamble_function_qualifier

* remove_instructions: insn_ids can be of type MatchExpressionBase

* ⚡

 improves performance of remove_instructions

* adds 'within' arg in extract_subst

- passes prefetch's within to extract_subst
- tests it

* fixup! test remove instructions

- modernize the tests

* fixup! adds 'within' arg in extract_subst

- modernize tests

* improves performance of remove_instructions

- look at the dep. graph and get a topo sort to avoid unnecessary
  operations.

* test that the prefetch was precisely targeted

* fixup! correct the function qualifier for Cuda target device functions

- use the good quotes

Co-authored-by: Kaushik Kulkarni <kaushikcfd@gmail.com>