Remove checking for parent / grandparent process

[soraxas]

This is just a proposal, where is it needed to check for parent & grand parent process (for cache)?

This is is different than the standard ssh-agent process where ssh-add ~/.ssh/id_xxx will adds the key to memory, and then it works in other terminals (I don't think it checks for parent process).

I understand that perhaps this approach allows for more fine-grain control of prompting request. However, it doesn't quite works right most of the time. E.g.

1. In plain ssh my_host the cache works well. But as soon as there are jump-host the cache no longer works
   • i.e., If your ~/.ssh/config has a host_b that jumps through host_a, it prompts for approval every time (probably because it spawns another ssh process, which makes parentProcess having different PID)
   • Reducing to JUST checking for grandParentProcess PID works, but then it fails again if we ssh to host_c via host_b via host_a, and so on.
2. My git pull does not cache as well, where in the goldwarden log it shows
   [INF] [...] [Goldwarden > SSH] >>> SSH Agent connection from fish>git>git
   so perhaps the git is using some subprocess, which causes checking of parent process fails
3. Again, even if we revert to checking for grandParentProcess, the above fails once we are using some wrapper process. I'm using yadm to manage my dotfile (which wraps git), and in goldwarden the process shows
   [INF] [...] [Goldwarden > SSH] >>> SSH Agent connection from yadm>git>git
   i.e., Checking for grandParentProcess will no longer cache the session.
4. Essentially, the intended cache logic can probably works if we walk the process hierarchical structure by getting their parents until we reach some interactive terminal, but I reckon that'd be too fragile.

[quexten]

Yeah, this was an experimental feature, but I've also found it unreliable, but don't know a better way to determine a program with an associated terminal or GUI cross platform. That's what I would have liked to cache by, and that is what 1Password does (I think).

I'm thinking of just moving to caching session wide to be honest, toggleable via an environment variable.

[quexten]

Great analysis and thank you for the PR!

[quexten]

^ I think the bypass should just be for ssh sessions for now, not vault access / pin entry. Other than that, I'm happy to merge.

[soraxas]

^ I think the bypass should just be for ssh sessions for now, not vault access / pin entry. Other than that, I'm happy to merge.

Yep the new commit should accounts for that