This means that a proposal of this kind, if targeting the Osmosis-chain, can immediately reactivate the IBC between the two chains?
No. Thank you for asking that question. It’s a really smart one. In case of Osmosis TFL made a
software change to the core software that actively and forcefully shuts down the channel grade connection between Terra and Osmosis on Blockchain level (or SDK level as I like to say)
In our analogy with the two participants in a phone call that is like taping the ears and mouth of the participants so they can’t talk anymore. But the speakers and wire were not touched.
In order to talk to Osmosis again we need to perform the following software update on the core software:
However, the “channel grade” connection between two chains is the more high level layer of IBC, meaning that a channel between two chains cannot exist without a underlying Tendermint client. But a Tendermint client does not need a channel to come into existence.
This means that we just go ahead making a governance proposal to unfreeze the Tendermint client that references Osmosis. But transferring funds would require GitHub PR 34 to happen (see link above)
My other question is why is this being proposed only now?
I don’t know exactly. I think educational-wise the community is pretty much unaware of the fact that MOST connections to other chains could simply be reactivated by the power of governance. There is no magical IBC switch that was turned off by TFL back in the days in May.
That’s one aim of the proposal: Teaching that the community has the power over IBC reactivation (Osmosis being a big exception). Not Terra Rebels or any other core developers do have the control over that.
What are the costs involved in maintaining an IBC channel?
You are raising the really good questions 
Costs are huge, in fact. As a relay operator you need to commit state mutating changes to both chains. That’s gonna cost gas fees. No one is gonna pay the relay operator for that. And me personally, I certainly do not have the bag to do so.
I made some experiments on my own transferring LUNC over a NEWLY created connection to Cosmoshub.
Every transaction was about .5 to 1$ in gas fees. Scale that up and you got the relay operating costs.
So in case of this proposal passes, there might be still a chance we have working Tendermint clients but nobody willing to operate a relay, because there is no incentive. That’s why I made another Agora proposal which was somehow spelled into the void:
After successful client update, I will certainly sit there and constantly update the Tendermint clients so they won’t expire (or in other words: I will hold the foot in the door so it doesn’t get closed anymore). But that’s actually not the way decentralization should work (me being a single point of failure)
We kinda have to wait for active relay operators to jump in to be able to see cross chain transactions again…