Can't tell if the problem is that the gas tube is not indexed with the carrier key, if there is a problem with a Bur on the entrance of the key at bottom of chamfer that is causing the divot, if the problem happened from a piece of blow primer~flake of brass finding it way between the key and the gas tube isntead, or if the problem is just the key center bore line not on the same plain as the center line of bolt isntead.

Myself since I do have a mill and indicators, would clamp the carrier with key in the mill on Center line of the carriers bolt channel passage, then using a rod, would indicate the center line of carrier key passage.  If the two are way off, easy enough to indicate off the top of the carrier flat for the Key to see if the problem is there, of if the problem is the key passage to it bottom flat off instead.

Problem in the upper carrier flat, B/C goes back for a replacement.  Problem in the key passage to bottom flat not parallel or bur at the entrance of the tube channel, walk in the part to clean that if its not too bad.

Being that the gas key was already replace, would bank that the problem is in the carrier itself, unless they just got a bad batch of keys that  have  the same problems isntead.

P.S. would help to know the manufacturer, since may be one that is just buying defective/import parts that are know problems to start with.

A, Make sure gas tube wiggle in the upper receiver through channel, and not being bound up by the barrel nut channel/spines.

B, Really looks like the center passage for the key is not on center line of the carrier channel.  
As stated, could be the flat on the top of the carrier for the key not square to center line of the carrier, or could be problems in the key milling instead.

Since this is a factory built upper, send it back again to have the problem resolved.

If this is something that you built yourself,  then should have center line barrel socket to center line of gas tube for the upper at .625", and this numbers should match for the key passage to  center line  of carrier ID passage as well.

Hence here for the B/C with it on center line of the carrier bolt passage, we  put a snug straight drill rod into the Key passage to run an indicator down the drill rod to make sure that the key passage in the holding fast at .625" (center line of the drill rod and will slight higher since we are measure the top of drill rod) from the center line of the carrier itself.

From your photo, either receiver is out of spec with the distance to center line not at .625", or problem with the carrier and key center lines not on .625" parallel bored centers isntead.

Fact is with the way the gas tube is begin rubbed, would dare to bank that the key channel passage is cocked upwards to begin with, and a easy fix.  The downside, when you remove the key to address problems with it, you have to use new gas key bolts to reinstall the key again.

And yes, we are well beyond that of a basic parts changer armor talent level at this point; where all the parts are QC'd before hand by someone else, to make sure what they are installing are in spec to begin with.

Only parts of the gas tube that should have wear, or even contact to the key's channel, is the end mushroom section at the end of the gas tube.

Could be upper receiver, with the center line of barrel socket and center line of gas tube channel not .625" as well.

Good catch, since I was just looking at the PDF drawing and pulled the wrong dimension with a quick glance: instead of pulling it up in cad to double checking the dimension instead.

Hence in the PDF drawing, its has the .625" measurement on the wrong side/not above the indicator arrows to indicate bottom of receiver to center line of barrel socket, and see why I pick up the wrong dimension at a glance isntead.

So as noted, .783 +/- .005 is correct.

Would need upper in hand to mic a few things to figure it out.

Hence on lock up,  gas tube is in the key with it end button baring on the ID of the key channel somewhat, the Front channel of the charging handle is outside the  key, and the Charging handle pulls the carrier back via the front top flat on the carrier in front of the key.  The front of the OD of the key is even reduced so you don't have binding of the front of the key to the ID of the charging handle as the charging handle comes back to grab the top front carrier flat to pull the carrier back.

So for the rubbing to take place on the gas tube by the key (changing handle front channel gets no where near the gas tube), the front of the carrier has to cam upwards, or the back of the carrier downwards during the initial CH pull, for the front of the key channel to get off the axis line of the gas tube to mark the gas tube on the way back.

Without mic' the upper, really can't answer the question.

Hell, the solution may be as easy as just re-tweaking the gas tube to resolve the problem itself.