Sorry but I don't get what that'd be in this case. Using fcurves.find()? I don't know what would be fast in this kind of situation.

Current loop looks at every f-curve for every sequence strip. if you have 100 strips and 100 f-curves, that's 10,000 comparisons.

fcurves.find() would be an improvement but that just moves the comparisons to C.

In this case you could do:

fcurve_map = {
    curve.data_path: curve
    for curve in fcurves 
    if curve.data_path.startswith("sequence_editor.sequences_all")
}

Then curve = fcurve_map.get(data_path) will give an efficient hash lookup.