Misleading error reporting in HTML repr

[jl-wynen]

When showing, e.g., a raw dataset, the HTML repr marks usedSoftware and inputDatasets as required and missing.

I misread the repr. The shown fields are marked with an asterisk because they are required by derived datasets. But they are only shown in the first place because they are set to not-None. So there are two issues here:

• Handle empty lists gracefully. See Scicat returns empty lists for list fields that were not set #226
• Clearly indicate what causes an error in the HTML repr. (What this issue is about)

I am unsure how to show the error type nicely.

• Tooltips don't seem to work well in Jupyter.
• There is no space to add an explanation. We could co-opt the 'Description' column and use a different font / colour to distinguish errors from regular descriptions. But that would hide the description.
• We could also use icons. But they would need a legend.
• We could add a new section of error messages at the bottom.

At least, we can omit the asterisk from fields that are not allowed to be used for a given dataset type.

See also #77