[Bug]: axes.prop_cycle behavior change

Nobody has claimed this yet.

PR #29808 (identified by `git bisect`) changed the behavior of:
```
$ python -c 'import matplotlib; print(matplotlib.rcParams.get("axes.prop_cycle").by_key()["color"])'
['#1f77b4', '#ff7f0e', '#2ca02c', '#d62728', '#9467bd', '#8c564b', '#e377c2', '#7f7f7f', '#bcbd22', '#17becf']
```
to now be a list of tuple of float rather than a list of str:
```
$ python -c 'import matplotlib; print(matplotlib.rcParams.get("axes.prop_cycle").by_key()["color"])'
[(0.12156862745098039, 0.4666666666666667, 0.7058823529411765), (1.0, 0.4980392156862745, 0.054901960784313725), (0.17254901960784313, 0.6274509803921569, 0.17254901960784313), (0.8392156862745098, 0.15294117647058825, 0.1568627450980392), (0.5803921568627451, 0.403921568627451, 0.7411764705882353), (0.5490196078431373, 0.33725490196078434, 0.29411764705882354), (0.8901960784313725, 0.4666666666666667, 0.7607843137254902), (0.4980392156862745, 0.4980392156862745, 0.4980392156862745), (0.7372549019607844, 0.7411764705882353, 0.13333333333333333), (0.09019607843137255, 0.7450980392156863, 0.8117647058823529)]
```
This is potentially problematic for end-users. MNE-Python we know the default `prop_cycle` (and everyone seems to use it!) so make use of the fact that we know `#7f7f7f` is in there, remove it, and create a new cycle. Our code now breaks with:
```
self.color_list.remove("#7f7f7f")
ValueError: list.remove(x): x not in list
```
And yes we should make our check better (e.g., remove any color sufficiently close to the bad color, etc.) but it seems like a potential regression that this behavior has changed in any case. Feel free to close if not!

Reproduce the two Python commands for rcParams.get("axes.prop_cycle").by_key()["color"] and inspect PR #29808, identified by git bisect, to trace the representation change. Compare the current and prior behavior, then add regression coverage for the expected public value and verify that the axes property cycle remains usable by downstream code.

Written by the indexing model from the issue text.