Rewriting BiGGId handling

[mephenor]

BiGGId parsing was rewritten as the old method had shortcomings when parsing ids with a large amount of underscores and pseudoreaction ids were handled wrongly, i.e. they got an "R_" prefix added that was not in accordance with the specification. The new method is regex based and while the major part of the rewrite is finished, some smaller issues remain.

• Ids that cannot be interpreted as BiGGIds are currently stored as abbreviations internally so they are not lost and corrected as much as possible to represent a valid SBase identifer, however identifiers starting with digits are not handled correctly yet -> prepend underscore?
• Invalid character replacement is done in a manner that produces ids that should match BiGGDB entries, however this needs to be verified by test cases
• Passing empty ids is not caught in all places, annotating species produces a "M_M" id across most (all?) models. Needs some investigation, as I haven't found the place where it happens yet. Maybe throw an exception and print the stacktrace when this happens for debugging
• Running the new code shows duplicate entries for some pseudoreactions in bigg_modelsdata, once with and once without the "R" prefix. I'll open a respective issue with them once I have the data for which models and reactions this occurs
• The new implementation is written in a way that we could actually make adding prefixes to ids a user choice, as proposed in #28 without much extra work
• In some models a compartment code prefix "C_" is found - should this be just stripped or added as an additional prefix that is not part of the specification?

[mephenor]

This should be done for now, might reopen if issues creep up after the beta release