New Session tests should handle more default browser capabilities

[gsnedders]

This ensures test_merge_platform_name and test_merge_browserName both
work even when a platformName and browserName capability is provided
in the default capabilities. And while we're fixing that, also fix the
case that pageLoadStrategy is provided.

[karlcow]

Not fully sure about the del statement there.

[whimboo]

Let's ensure that each test only verifies the specific behavior indicated by its name.

[gsnedders]

Let's ensure that each test only verifies the specific behavior indicated by its name.

I presume these use pageLoadStrategy to be able to ensure that firstMatch is behaving correctly — that browserName/platformName in firstMatch isn't just being totally ignored, as otherwise they would pass. This seems to go all the way back to when @jgraham originally wrote the tests in ec5cd8c.

We certainly can remove pageLoadStrategy, but then it becomes indistinguishable which firstMatch is being applied (if any!), and when the test is supposedly testing the merging, that seems important?

[whimboo]

Ah, yes. Thank you for the explanation. So lets apply the following suggestion to the relevant tests to make it clearer.