Consider using a predictable regex library

[mdoering]

There are problematic cases when the name parser takes very, very long to run some regular expressions.
It is a well known problem with backreferences in particular and the use of NFA. E.g. see also https://bugs.openjdk.org/browse/JDK-8260688

There are alternative regex implementations using DFA, but all have less features that are currently used by the name parser. But maybe these can be replaced and we can use a deterministic engine and do aways with parsing timeouts?

Alternatives:

• https://github.com/google/re2j port of https://github.com/google/re2/wiki/WhyRE2
• https://github.com/openresty/sregex

Related issues:

• Parsing timeout for long authorship: de Puytorac, ... #12
• Handle all IPNI names, often infragenerics #15
• parsing timeout when using gbif-parse  globalbioticinteractions/nomer#89