I like this idea. It's basically a "prepaid future block space". This could even be generalized, i.e. not only usable for Lightning channels, but also for all kinds of transactions - you anyway couldn't avoid it to be used for other purposes. I have not heard about such a proposal at this moment, although it's still possible somebody has proposed that.

There's only one problem I see: When transactions are cheap, it is likely that some entities could reserve lots of space and then make block space unnecessarily scarce, for example exchanges who want space for future consolidation transactions. This would however also drive up the fee, like if normal transactions were stored instead, and thus it could not be abused infinitely. In general it would lead probably to a more balanced fee marked, with less low-fee and less high-fee periods.

Some time ago I had a somewhat related idea: change the "weight formula" of some transaction types, to make simple payments and Lightning closure transactions occupying less weight than normal transactions. This would make channel closures in congested times easier. The problem is however a bit similar to the one of your idea: if you lower the weight too much, then people will find ways to use it for other purposes, maybe even for storing data in some way (see Stampchain's or Doginals' approach) and then you have effectively a block size increase.

Your proposal is better than this one, because there would be no block size increase even if the mechanism was gamed somewhat, but there could be still the problem of excessive "space reservations".

I don't think that a hard fork would be achievable for this on Bitcoin, at least outside of an emergency situation. You could try however to find an altcoin project which is more accustomed to hardforking, and in the case of the mechanism working well it could maybe taken under consideration.