timing(LoadUnit): adjust rar/raw query valid generate logic

[cz4e]

LoadQueueRAR and LoadQueueRAW need to use the valid signal of the query to select the index of the queue at load unit s2 from FreeList (because of FreeList set enablePreAlloc). It can put all conditions except fast replay in s3 for judgment (fast replay can fix #4149). If other conditions are setted in s3, LoadQueueRAR and LoadQueueRAW can revoke the corresponding entry. This is timing friendly, can eliminate 4 levels of logic for query valid generate, and does not add additional logic in s3

[linjuanZ]

More specific description for the modification, please.

[XiangShanRobot]

[Generated by IPC robot]
commit: d2e97bd

commit	astar	copy_and_run	coremark	gcc	gromacs	hmmer-Vector	lbm	linux	mcf	microbench	milc	namd	povray	wrf	xalancbmk
d2e97bd	1.806	0.442	2.635	1.232	2.155	1.507	2.141	2.362	0.937	1.401	1.949	3.115	2.534	2.269	3.302

master branch:

commit	astar	copy_and_run	coremark	gcc	gromacs	hmmer-Vector	lbm	linux	mcf	microbench	milc	namd	povray	wrf	xalancbmk
6f9d483			2.636	1.221		1.500	2.145		0.942	1.405	1.998	3.113			3.303
602b407			2.636	1.230		1.500			0.943	1.408					3.303
6520f4f	1.819	0.442	2.636	1.222	2.143	1.500	2.145	2.360	0.943	1.405	1.998	3.113	2.565	2.271	3.303
1abade5	1.819	0.442	2.636	1.237	2.143	1.500	2.145	2.360	0.943	1.401	1.998	3.113	2.565	2.271	3.303
881e32f	1.819	0.442	2.636	1.229	2.143	1.500	2.145	2.360	0.943	1.408	1.998	3.113	2.565	2.271	3.303
59ef52f	1.819	0.442	2.636	1.234	2.143	1.501	2.140	2.361	0.939	1.405	1.984	3.113	2.565	2.271	3.303
16c2d8b	1.819	0.442	2.636	1.228	2.143	1.501	2.140	2.361	0.939	1.405	1.984	3.113	2.565	2.271	3.303
0d1fb64	1.819	0.442	2.636	1.216	2.143	1.501	2.140	2.361	0.939	1.401	1.984	3.113	2.565	2.271	3.303
fa5e530	1.819	0.442	2.636	1.233	2.143	1.501	2.140	2.361	0.939	1.405	1.984	3.113	2.565	2.271	3.303
bf4beb4	1.819	0.442	2.636	1.228	2.143	1.501	2.140	2.361	0.939	1.401	1.984	3.113	2.565	2.271	3.303