Intra/inter reference area versus performance

[iAvoe]

From the original document x265.readthedocs.io and a bunch of paper, I've found that both x264/5 are only using one row of macroblock/ctus for inter and intra referencing (or just inter, not very sure about intra). And in x265, frame-thread can be used to make the whole inter frame(s) available but in cost of adding delays.

My questions are:
1.if I'm right, could there be an option to expand reference area by waiting for more reference blocks outside are generated, if blocks within it are not good enough (causing error)

2.does lookahead-thread option increases performance when frame-thread is disabled

3.is inter frame referencing not as relevent as I think, because by increasing mode decision / loop filter costs (things to do before generating reference blocks) are giving better compression

4.is the whole intra frame available to reference at very first

Thanks for reading though!