Improve Gracefulness of RIS Import

[ursjoss]

Based on suggestions provided by @adamsmd in issue #129, I'd like to improve the way KRis handles the import.

While the apparent issue of unexpected types of the RIS Tag M1 (containing ranges or otherwise non-numeric content) was already worked around in PR #133 by changing the Type of the property associated with Tag M1 from Long? to String?, the user experience of importing from RIS files is not very good. (Side note: This is most likely the case due to the fact that I only use KRis for exporting, not for importing. Hence the import is less polished by the lack of "dog-fooding").

As a general direction, KRis should not only keep track of the type safe value of the tag content, but also of the raw value. There may be other types that may contain content not complying with KRis' assumption of the data type. In case the type conversion fails, the user of KRis should be

• informed about what went wrong (which tag in which line of the KRis input was causing the issue)
• able to fall back to the untyped raw value.