Speedup CI builds

### Description

currently macOs builds are ~ 2x slower than linux or windows builds.

This is caused by macOS builds being universal builds, requiring double the build time versus a single separated arch (x86 or arm).

- Do we need universal builds for every dev branches?

Fix would be to change ci runner config to trigger in-parallel separated arch build using runner matrix configuration and an extra `lipo` step could be added to glue together into universal build if required.

## IMPORTANT
this task is only about speeding up **development builds (PR builds)** it shouldn't change anything about the currently working nightly and stable build / production builds