Consider making the GuardMgr mandatory everywhere

2024-06-11 14:28:48	gabi	arti-folks: why is the guardmgr (usually) optional in tor-circmgr? I noticed the path building logic is subtly different when it's set to None (e.g. we apply fewer restrictions when selecting the guard). Does anyone know why that is?
2024-06-11 14:29:28	+nickm	gabi: I think for historical reasons.  IMO it might be a good idea to make it mandatory.
2024-06-11 14:30:13	gabi	I see. I'll open a ticket
2024-06-11 14:30:15	+nickm	dgoulet: IYO is there any reason for guards to ever be optional?  IMO there isn't a good reason.
2024-06-11 14:30:48	+nickm	gabi: there could be tests that depend on an absent guardmgr I guess?  But those tests should be rewritten IMO
2024-06-11 14:30:48	+dgoulet	C-tor has this "don't use guard" option iirc... but not sure it is greatest idea to have lol
2024-06-11 14:31:07	+nickm	yeah, and I would not be surprised if it were one of those options where everything breaks when you turn it on
2024-06-11 14:31:34	+dgoulet	yeah I lean towards not having such option in arti for now ...
2024-06-11 14:32:03	gabi	yeah, if the are no guards, I think we are slightly too lenient when selecting nodes AFAICT...
2024-06-11 14:32:21	gabi	But those tests should be rewritten IMO> I agree, and they should be easy to rewrite (I've just rewritten some of mine)