Improve function name distinctions

There are several methods that appear in different logical parts of the code (e.g., the semidiscretizations, the DG solver, a callback etc.) but are named identically. At least for some of them it is not (immediately) clear, why they shouldn't be different functions such that it is easier to distinguish them when reading the code with a "dumb" editor.

Since this is something I encountered before, and now @martin-pmf has brought up again, I thought to start collecting possible functions that could benefit from splitting it in multiple functions (feel free to edit and amend):
* `create_cache`:
* `create_cache_solver`
* `create_cache_indicator`
* `create_cache_controller`
* `create_cache_analysis`
* `compute_coefficients(!)`
* `compute_coefficients_semi`
* `compute_coefficients_solver`
* `calc_error_norms`:
* `calc_error_norms_semi`
* `calc_error_norms_solver`

Search the codebase for create_cache, compute_coefficients(!), and calc_error_norms, then group their call sites by semidiscretization, DG solver, callback, controller, or analysis context. Review each group's callers before deciding the distinct names; done means the selected functions and their references consistently use names that reveal their logical role.

Written by the indexing model from the issue text.