Werk #4002: Speed up display of BI aggregations

Previously, each Apache process had to compile the all BI aggregation trees by itself. This generated lots of CPU load during core restart, especially on larger BI configurations.

With this update, an Apache process no longer computes the BI trees only for its own use. The computation results are now shared between the different processes through cache files.

If you encounter any bugs with this new mechanism you can switch back to the old implementation via the global settings option Use legacy compilation for BI aggregations (slower).

To the list of all Werks