olevba: test if Aho-Corasick is applicable and faster than simple string search

[decalage2]

olevba looks for many strings in VBA source code, XLM, and P-code. The Aho-Corasick algorithm is meant for that, so it might be faster than the current implementation, which simply searches strings one by one.
However, olevba uses regex for the searches to find word boundaries, so maybe Aho-Corasick cannot be used as-is.
Potential candidates:

• https://github.com/JanFan/py-aho-corasick
• https://github.com/WojciechMula/pyahocorasick

On the other hand, for now olevba search looks fast enough most of the time, so this would only be needed if we experience performance issues for some samples, or if someone uses olevba to process many samples at large scale.