#218 Improve cache access for runtime

[Erik Hänel]

If one accesses many different caches or many different regions of one cache in equations inside a loop, the runtime is overproportional increasing compared to the same operations done only with scalar values. Find a solution for caching the relevant parsing steps instead of re-calculating them every time.

Analysis:
This can only be done with multiple changes:

• The current way of accessing data objects has to be changed so that the index evaluation is centralized in parser_getIndices(). This way, all optimizations of the index evaluation are distributed to all relevant locations.
• Furthermore the bytecode caching in loops has to be adapted so that the index evaluations are stored as well. It's possible that this change will require also a change in how the bytecode is cached in principle, e.g. changing the cache from a 1D cache array to a 2D one.

Implementation:
Change was implemented as proposed by the analysis. In addition, multiple changes were done in Loop::calc() so that the function after the first iteration knows, which of its parts is necessary for the current equation. A runtime improvement up to a factor 4 was achieved. Implementation test were passed successfully.

Documentation:
Not needed. Internal change without a change in functionality.

Tests:
Durability tests returned no deviations. Change implemented successfully.