fix(mmu): adjusting the timing of non-virt PBMT assignments

[Anzooooo]

We have already assigned a value to cpu.pbmt at the end of gpa_stage. It should not be reassigned, as this will cause it to be overwritten incorrectly:

NEMU/src/isa/riscv64/system/mmu.c

Lines 371 to 382 in 112f2fd

	}
	word_t pg_mask = ((1ull << SVNAPOTSHFT) - 1);
	pg_base = (pg_base & ~pg_mask) | (gpaddr & pg_mask & ~PGMASK);
	}
	cpu.pbmt = pte.pbmt;
	return pg_base | (gpaddr & PAGE_MASK);
	}
	}
	raise_guest_excep(gpaddr, vaddr, trap_type, is_support_vs);
	return gpaddr;
	}
	#endif // CONFIG_RVH

[github-actions]

NEMU Performance Results

Test	Instructions Executed	Estimated Host Throughput (instr/s)	Actual NEMU Throughput (instr/s)
bitmanip.bin	5.147e+07	2.691e+07	2.495e+07
coremark-riscv64-xs-rv64gc-o2.bin	1.731e+08	1.938e+08	1.732e+08
coremark-riscv64-xs-rv64gc-o3.bin	1.726e+08	1.967e+08	1.699e+08
coremark-riscv64-xs-rv64gcb-o3.bin	1.697e+08	1.788e+08	1.555e+08
amtest-riscv64-xs.bin	8.673e+06	1.830e+07	1.164e+07
aliastest-riscv64-xs.bin	7.697e+06	1.788e+06	2.814e+06
softprefetchtest-riscv64-xs.bin	7.744e+06	3.413e+06	5.600e+06
zacas-riscv64-xs.bin	1.212e+07	5.340e+07	1.439e+07
linux-hello	1.854e+10	4.069e+07	4.609e+07

• Host throughput is estimated based on 4GHz CPU frequency and IPC=2.5.
• Actual throughput may vary based on the host CPU performance.

[cebarobot]

LGTM