Support latest graph_scheduler about mdf

[kmantel]

There's a new check for invalid termination condition input, and the two failing tests look to be hitting that. It looks like they don't specify a time scale for the termination condition:

MDF/examples/ACT-R/count.json

Lines 413 to 422 in 2e9cd9c

	"termination": {
	"check_term_true": {
	"type": "JustRan",
	"kwargs": {
	"dependencies": [
	"check_termination"
	]
	}
	}
	}

A correct example:

MDF/examples/PsyNeuLink/SimpleFN.json

Lines 15 to 27 in 2e9cd9c

	"termination": {
	"environment_sequence": {
	"type": "Never",
	"args": {}
	},
	"environment_state_update": {
	"type": "AllHaveRun",
	"args": {
	"dependencies": []
	}
	}
	}
	},

In prior graph-scheduler versions, that condition would have been silently ignored.

This line looks like the culprit:

MDF/src/modeci_mdf/interfaces/actr/importer.py

Line 231 in 2e9cd9c

termination={"check_term_true": cond_term},

[kmantel]

I'm not sure the exact intent of the model, but my guess would be that

 termination={"environment_state_update": cond_term}, 

would be the replacement to make.

Maybe @davidt0x has input

[pgleeson]

@davidt0x Did you suggest in our last call that your latest PR solved this issue?

[davidt0x]

I did suggest that but I forgot to include the fix, will generate another PR, sorry for the confusion.

[davidt0x]

This issue is fixed with PR #524