I wrote a crude version of a similar workflow back when our company first started large adoption of Revit and we needed to collect and categorize all of our floating details into a more organized detail library.
The issue that I was running into with just node usage (no python experience) was the sheer number of potential elements that required searching. Our details usually float around 10+ text notes, and we were sifting through several thousand. The script winds up needing to inspect an large number of Text Notes, and it sounds like from your workflow that you'd be incorporating string comparisons which means inspecting the substrings inside the Text Notes as well...all of this is far easier if your company uses Detail Items which are parameterized to construct your details but that does not appear to be a popular practice.
There's also the issue of querying open models versus a library of models - the Orchid package exists and allows for processing through multiple models, but that's just another level of multiplication.
If you have a details library established, it might be a useful project workflow at deliverables to kick the details back out to a 'holding area' inside the details library where they can be compared against similar details for modification and then ultimately categorized and kept/eliminated.