Metapool project is a superset of “fallback” approach, allowing to spread the funds of delegators among the whitelisted or governed pools.
I don’t think direct fallback approach make that much sense because users already struggle with choosing a validator, so choosing two or three validators as fallback is even harder.
Generally, I think allowing anyone with at least 5k $NEAR becoming in the active set and starting to receive rewards should be sufficient given current NEAR price (as price appreciates we can later consider reducing this threshold lower via a protocol upgrade). Others can then delegate to these validator as soon as they are slotted to be active validators.
Metapool approach could then help with decentralizing the delegations more to these new validators on behalf of users.
There will be also more ways to come by $NEAR via lending soon or a validator bootstrapping DAO can vote to stake initial 5k to applicant and hold it there until they receive their own delegations.
Ideally, with this approach we should aim to have at least 1k active validators. Obviously still majority of stake will be parked with top validators, but having low threshold will allow projects to become validators and operate from the rewards (see discussion here: NEAR Ecosystem Treasury DAO - #39 by galaxyonline).