Improve documentation of `addcorpus.models.Field.language`

The [`language` field of the `Field` model](https://github.com/UUDigitalHumanitieslab/I-analyzer/blob/fb1d88c068bc7e8a74372baa0c6951c1eba90c14/backend/addcorpus/models.py#L354) looks strange to me:
It has a `CharField` with `blank=True, null=False`. So then we should set a default, I believe? NB `null=False` is not necessary to state explicitly (and it's not advised to set this to `True` on `CharField`s anyway). `validate_field_language` seems to assume that fields which don't have an explicit language tag set would have to have "dynamic" set. I'm not sure I agree with enforcing that every field, or for "dynamic", the corpus, has a valid language tag, in some corpora we simply don't have that kind of metadata. A docstring in `validate_field_language` about the meaning of the "dynamic" setting may also not be misplaced.

Start with backend/addcorpus/models.py, especially Field.language and validate_field_language. Review the current field defaults and validation behavior against the questions about missing language metadata and the “dynamic” setting. Done requires an agreed decision on the intended behavior and corresponding documentation, with validation expectations clarified.

Written by the indexing model from the issue text.