Transitioning to an XML, Namelist, JSON, or other format for parameter file

[rgknox]

While netcdf has its benefits, there are also benefits to storing parameter constants in other formats, such as XML, fortran namelist, json, or others. Use this space to plan this change.

Format	Pros	Cons
Netcdf	fast binary/parallel IO	requires compilation step
	error checking	conda has weak support
	strong fortran support
XML	high conda/python support	too verbose (lots of tags)
	simple python interpretation	not strong fortran support
		slower IO (text)
Namelist	easy fortran implementation	requires new python scripting
	simple visual format	units/dimensions/etc would be comment only
JSON	like XML but less verbose	less ubiquitous than xml, maybe less python support

Here is a design document that I started wtih @glemieux :

https://docs.google.com/document/d/1ufKlQzQAtBQLFwICfnv59Kp7QslgdUT2CIfuh0D5mIM/edit?usp=sharing

[rgknox]

Note this is not part of the CLM6 milestone set, this is a longer term design plan (if implemented at all).

[glemieux]

All: Please use this as a "parent" issue to create sub-issues that work to address this larger task.

[rosiealice]

Are there any examples, e.g. from other components, of what naelist and xml parameter files might look like? I feel like that might help us figure out whether our use case suits one or the other better. The most pressing thing is that our parameters can have multiple dimensions, and the format needs to be able to support that in a non-riduculous way :)