I’d say improve the data, move with your life, and if it’s still interesting to you in six months, do it again.
In the mean time, 2000 opening_hours tags will be easier to parse.
Other problems with the specification or the editing tools or the parsers should be fixed too but no one has volunteered. For now, @matheusgomesms is willing to improve the data right now, so I say go for it.