Each different engine probably has it's own optimum lean burning AFR in the 15-17:1 range based on a whole bunch of different factors (mostly engine design, although it may change with compression and plenty of other things).
If you're tuning the tables to provide a constant AFR across the board at low loads anyway, you'll end up with essentially the same thing. Keeping it locked in open loop, you could theoretically find the optimum AFR at different part-throttle loads. But without spending the time to actually tune it to your engine, there won't be any consistent, repeatable advantage. Engine performance is does not vary that much on hair-splitting differences in AFR. The only real advantage of locking in open loop is having finer control over the timing curves, which are much more influential than A/F ratios (MBT or mean best timing still applies regardless of the engine's operating load and conditions).
It shouldn't be too much of a problem on closed to open loop and vice versa - I had initially thought it would be, but our ECU already has "tip-in" enrichment and leans out on "lift-off" based on the TPS, and the fuel dumped in under load is quite a lot compared to even the fuel used under cruise at 14.7:1. It has this feature mostly because of the fact that in a MAF based system, there *is* a delay in airflow detection from the sensor to the throttle body. To compensate for it, the ECU adds or subtracts fuel based on the rate at which the TPS voltage changes. Coincidentally, even though a blowthrough setup is supposed to have better response, since you can't fiddle with the tip in, it will overcompensate when you mash the throttle, or lift off rapidly.
Fuel trims should work fine too, since all they're doing is trying to flow the right amount of fuel to reach a certain A/F ratio. The computer trims are dumb and don't care what mixture it's targeting - all it cares about is that it flows the right amount of fuel to make the O2 sensor happy and hit the right A/F ratio. Since the WB can simulate a narrowband at all, it should work fine.
Trivia section:
It is interesting to note that a narrowband oxygen sensor is actually a miniature fuel cell that burns CO and O2 to produce CO2. It generates an electrical voltage that corresponds to the amount of O2 and CO2 in the exhaust stream, which incidentally can be corresponded to AFR. But it's not actually a linear relationship, which is precisely why it is accurate only around the point it is tuned for, where it outputs 0.45 volts (which is not quite 14.7:1, but it's close enough in that range).
A "wideband" sensor is not actually a sensor that somehow detects differing gas concentrations directly. What it actually does is that it uses a normal narrowband sensor, but instead of simply exposing it to free exhaust stream, it uses a fancy electronic device called an "ion pump" to either restrict or pressurize flow to the narrowband sensor to keep it's output in the .45 lambda range. Since the sensor only absorbs oxygen and CO2, it essentially fakes a stoichimetric ratio at the sensor.
The way it actually "senses" is that the current draw of the pump required to "bump" the voltage output of the sensor to the correct value can be mapped to determine the A/F ratio of the gas that the pump draws from. Even though it's not linear, it's repeatable. (not a big problem for electronics - sensors like thermocouples, MAP sensors and mass-airflow sensors are also non-linear, but also have repeatable outputs)
The wideband requires a "controller" that in a way acts just like the closed loop system in your engine - if the sensor is too lean, it increases the pump flow, and if it's too rich, it restricts it. Instead of applying "fuel trims" to correct it's behavior, it keeps track of what the fuel trim would "need to be" and uses that to report the voltage to the gauge,logger, etc. Coincidentally, it's because of this that it has to be calibrated every so often. It cannot "trim" its own control system.
