Add QM greedy cover option and enhance CUDA telemetry in synthesis tools

- Introduced a new command-line option `--cuda-qm-no-host-cov` to enable QM greedy cover functionality, allowing the covariance matrix to be maintained on the GPU for improved performance.
- Updated `verilog2penl.cpp` and `verilog2plsav.cpp` to integrate the new option, enhancing user control over synthesis behavior.
- Enhanced timing reporting by including elapsed time for each synthesis pass, providing better insights into performance metrics during execution.
- Improved CUDA function signatures and added new kernels in `pe_synth_cuda_u64_cones.cu` to support the updated QM functionality, optimizing resource management and performance tracking.

Author: MacroModel