Further speed improvements

Problem/Motivation

This is a follow-up based on #3007091: Performance issue in ConfigurableResourceType. This aims to add further speed-up on the method, by pre-generating the whole set of mappings in a greedy manner instead of 1 by 1.

Proposed resolution

Instead of lazy initialization of cache on miss - have all the mappings initialized on the first miss.

Remaining tasks

Changes on the method internals.

User interface changes

None.

API changes

None.

Data model changes

None. Maybe cache storage structure.