[fpv] prim and top-level IP improvements

### Description

For Earlgrey-PROD.M4 and as part of #22721 @rswarbrick managed to substantially improve FPV pass rates and coverage for the prim and top-level IP regressions:
- https://reports.opentitan.org/hw/top_earlgrey/formal/ip/summary/latest/report.html
- https://reports.opentitan.org/hw/top_earlgrey/formal/prim/summary/latest/report.html

As documented by @rswarbrick [here](https://github.com/lowRISC/opentitan/issues/22721#issuecomment-2126740422), the following two things remain to be resolved for future releases:
- Improving pass rates for `prim_fifo_sync_fpv` and `prim_fifo_async_sram_adapter_fpv`. These might need some tooling workarounds.
- Improving coverage numbers for `pinmux` and `rv_plic`. Most existing gaps are due to COIs of SVAs not touching the signals of interest, in particular inside `u_reg`.

Both these items are of lower priority but if we decide to investigate this further, this issue can serve as a starting point / reference to previous work.

Start with the work and notes referenced in #22721, then review the formal reports for prim_fifo_sync_fpv, prim_fifo_async_sram_adapter_fpv, pinmux, and rv_plic. Investigate the stated tooling and COI gaps, especially around u_reg; done means improved pass rates or coverage for the affected regressions.

Written by the indexing model from the issue text.