Obtain timeseries step information from dims or observed

[ricardoV94]

Follow up to #5690 so that steps can also be retrieved from observed and dims when possible.

Other changes:

• Renamed pandas_to_array to convert_observed_data
• Made AR steps more consistent with expectation / GRW

[codecov]

Codecov Report

Merging #5743 (504da82) into main (c76b9b9) will decrease coverage by 0.08%.
The diff coverage is 94.54%.

@@            Coverage Diff             @@
##             main    #5743      +/-   ##
==========================================
- Coverage   89.22%   89.13%   -0.09%     
==========================================
  Files          75       75              
  Lines       13824    13844      +20     
==========================================
+ Hits        12334    12340       +6     
- Misses       1490     1504      +14     

Impacted Files	Coverage Δ
pymc/distributions/timeseries.py	77.55% <93.75%> (+1.27%)	⬆️
pymc/aesaraf.py	91.46% <100.00%> (ø)
pymc/data.py	79.83% <100.00%> (-3.79%)	⬇️
pymc/distributions/shape_utils.py	97.07% <100.00%> (ø)
pymc/model.py	86.00% <100.00%> (-0.55%)	⬇️

[ricardoV94]

We really need to sort out what's going with #5739, it's failing nearly every time

[michaelosthege]

Looks really good!
I particularly like the test cases.

[michaelosthege]

Can you add a comment what this line does?

[ricardoV94]

I think the function self documents itself well enough:

pymc/pymc/util.py

Lines 337 to 351 in 02897d1

	def check_dist_not_registered(dist, model=None):
	"""Check that a dist is not registered in the model already"""
	from pymc.model import modelcontext
	try:
	model = modelcontext(None)
	except TypeError:
	pass
	else:
	if dist in model.basic_RVs:
	raise ValueError(
	f"The dist {dist} was already registered in the current model.\n"
	f"You should use an unregistered (unnamed) distribution created via "
	f"the `.dist()` API instead, such as:\n`dist=pm.Normal.dist(0, 1)`"
	)