Unduplicate frida_libpng fuzzer code using FridaRuntiemVec

[jejuisland87654]

making most of the code in frida_libpng/fuzzer.rs shared using FridaRuntimeVec, reduced the file from 494 lines to 274

[domenukk]

I wonder if the performance is different at all, since now there's dyn dispatch. I guess it shouldn't matter in practice, though

[jejuisland87654]

good point, is there a convenient way to measure?

[jejuisland87654]

also a possible way to tackle this is to have get_first_asan_runtime, get_first_coverage_runtime, ... and keep these objects indices instead of calling match_first_type every time

[domenukk]

An alternative option to this is #2792.
Imho no matter on what option we settle (for the fuzzer.rs), we should definitely still merge the FridaRuntimeVec part of this PR. I'm a big fan of dyn dispatch for ease of use.