Here is my take. Some things I'm sure about, but there are a few things I am unsure about.
What I am sure about is that a float of 3.40V/cell works just fine, I use this all the time to provide load-support without wasting battery resources. The question comes down to what causes more wear? Floating the battery at 3.40V with almost no current going in or out, or allowing the load to draw the battery down and push it into bulk cycling a few times every day? I think floating (load support) is the better choice.
But now there is the question of load support. At 3.40V/cell, a modest load on the system will draw the battery down and if it goes south of around 3.35V then it will start pulling current from the battery. So the question is, is 0.05V a sufficient differential for the power source to be able to supply load support? In many cases the answer is NO, it is not. Heavy loads will begin to draw down the battery even if there is plenty of external power available. So a better 'float' voltage for load support is 3.45V/cell, not 3.40V/cell. With the float set at 3.45V/cell, a nominal load takes it down to roughly 3.42V/cell and a heavy load takes it down 3.35 to 3.375V/cell... still high enough that it doesn't pull power from the battery.
So in a 'UPS' style application I prefer setting the load support float to 3.45V/cell.
But that is where I become unsure. Is it ok to leave 3.45V/cell on a LiFePO4 battery indefinitely? That is what I don't know the answer to.
At the moment I am doing this on two Bluetti EB55's... I am driving the adapter input with a 24V current-limited power supply (limited to around 7.2A or 175W). The actual supply output is 24.2V (3.457V/cell, the EB55 is 7s I think). But the system has a continuous load of around 100W and under load I measure around 24.02V on the adapter input. So this would really be a cell voltage of 3.43V. I am fairly confident that 3.43V/cell is 'ok' to leave on 24x7x365 days a year. But I am not 100% sure.
Obviously holding higher cell voltages such as 3.50V or 3.55V/cell is a bad idea. Similarly, holding 3.40V/cell should be fine. The question is... what about slightly higher voltages in the 3.41-3.45V/cell range?
There is also the question of the BMS not balancing the cells at those voltages, but LiFePO4 holds its balance pretty well so in my case I just do a full charge every 6 months or so (that's the plan anyway) to keep the cells in balance.
--
There is a similar argument for non-continuous / non-UPS use. If you have a solar-only system and want to take advantage of load support, the solar system will be charging the pack at 0.2C or lower (most likely) so there is no need to set a bulk voltage higher than 3.50V or so as long as the BMS does some balancing. We again want a float voltage that will properly load-support the pack. And again the appropriate range is 3.40V to 3.45V. In this situation due to the daily cycling I don't see any problem with floating at 3.45V from the solar system.
One thing I do take issue with is trying to bulk or float at voltages below 3.40V. If the battery is mostly full, then bulking to 3.35V to 3.40V will do a reasonable job keeping it full. But if the battery becomes depleted then bulking to 3.35V or 3.40V is going to leave a LOT of solar power on the table. The battery just won't get charged up by the solar system! The solar system will limit the current too much. So while a bulk of 3.40V might appear to be fine, it is NOT fine for charging a depleted battery pack. It is not a good target voltage for the Bulk stage.