Add EquationSystems read/write capability via istream/ostream

[loganharbour]

No description provided.

[roystgnr]

Fingers crossed on the CI, but I was hoping it would be this simple!

[roystgnr]

The finger-crossing thing never works.

[lindsayad]

Is this ultimately for idaholab/moose#21011?

[loganharbour]

Is this ultimately for idaholab/moose#21011?

Yep. Gonna use libMesh for the system loads via the moose restart system.

[loganharbour]

Good now. Gonna add unit tests and clean up doxygen.

There's some real ugly std::function use here... couldn't use a templated lambda because we don't put the methods in the headers. It's a unique case when you need to show how to load the per-processor data, without actually loading it (because it might not exist)

[moosebuild]

Job Coverage on 92ce4b7 wanted to post the following:

Coverage

	aef87c	#3566 92ce4b
	Total	Total	+/-	New
Rate	60.23%	60.23%	+0.00%	72.41%
Hits	49373	49383	+10	21
Misses	32605	32605	-	8

Diff coverage report

Full coverage report

Warnings

• New new line coverage rate 72.41% is less than the suggested 90.0%

This comment will be updated on new commits.

[loganharbour]

Ready.

[roystgnr]

I still like it, assuming John doesn't object to the technically-an-API-change or to us getting rid of that hack for reading our (third-most?) ancient file format with infinite elements.

[jwpeterson]

I don't really recall exactly the details here, but we aren't dropping support for reading xda/r files with infinite elements, just for really old versions of xda/r files.

[roystgnr]

That's correct; only versions so old that we didn't put that info in the header should be affected.