DFT atomate workflow and input generation

[brookwander]

Use cases:

1. Enumerate DFT inputs for adslabs without running inference
2. Enumerate DFT inputs for adslabs at the end of inference
3. Run DFT on outputs of (1), (2)

Scope:

What will be done:

1. Add section to config file for DFT settings
   a. Add flag to generate DFT inputs:
   - "unfiltered": generate inputs for all adslabs
   - "filtered only": generate inputs for lowest-energy configurations following inference
2. Generate DFT inputs from catlas outputs
   3. Add section to config file for mongodb, vasp files, etc. to run DFT
   a. atomate2-based relaxations based on output of (2)
   4. Custodian incorporation

What will not be done:

1. Run automated DFT in the background on laikapack
2. Feedback
   Execution plan:
   This should be detailed, you should consider the infrastructure currently in catlas. What specific functions need changing? Are there several significantly different alternatives to implement? What are some advantages and disadvantages to different approaches
   You should also use this section to delegate tasks. Upfront tasks should be outlined, and then you can work with your subteam to delegate them.

Task	Owner	Due by	Dependent?
DFT config mod	Kirby	Content Cell	Content Cell
DFT input generation	Kirby	Content Cell	Content Cell

Consider any problems / risk:
A lot of opportunities to leverage the team here. This upfront planning will allow us to prioritize things better and also prevent wasteful work.