[ENHANCEMENT] more informative error when no setting is provided

[mbakker7]

Pastas can not do certain tasks if no settings are provided in the StressModel. For example, when the series is not long enough for the provided warmup, I get the following error:
ValueError: Simulation contains Nan-values. Check stresses time series settings!.

This error is not informative. The series is not long enough and needs to be extended in the past. I suggest to state that no settings are provided to the StressModel so that pastas doesn't know what to do. With instructions on how to do that. Also, there may be easier ways to specify the settings.

[martinvonk]

Maybe make providing StressModel settings mandatory? Even though an input time series is long enough.

[martinvonk]

fill_before and fill_after function should not fill nans when settings are not provided. Raise the error there.

[martinvonk]

Related to the time series settings.

I found out about there exists something like the TypedDict in Python. This can help us to be more informative about the settings have to be provided. For stresses the the following settings have to be known: 'sample_up', 'sample_down', 'fill_nan', 'fill_before', 'fill_after'. With a TypedDict users get syntax highlighting if these settings are not provided. Small change but maybe very useful.

So code would look something like this:

from typing import TypedDict

class StressSettings(TypedDict):
   sample_up: str
   sample_down: str
   fill_nan: str | float
   fill_before: str | float
   fill_after: str | float

[dbrakenhoff]

Closed by #639