here is a graph of your wideband during closed loop, i threw injector pulsewidth in there too so you can see that the lean points are actually DFCO (injectors off during decel)16846

here, this is what a good wideband analysis of narrowband looks like. (this is your data and wideband settings)16845

oh and you have a lot of knock at 3-4k wide open throttle, you should look into that

everything seems fine except that you might be trying to analyze wideband AFR in closed loop (which is a fairly useless analysis) and trying to do a percentage deviation without the AFR target patch...

lets see a log