I think after as many landing successes they have figured out that quick return with minimal engine burn then a well controlled landing with landing legs deployed at the last seconds. That might be one of the quickest returns to the pad flights yet. Well orchestrated.
Some of the at-sea landings have taken as long as 9.5 minutes to land, IIRC. The ones with heavier payloads that have left the Falcon with not much fuel and no "partial boost back" burn. The current version of the Falcon is "Full Thrust", which now may also be considered "Block 4" (as they have a Block 5 version coming out in the Fall, and keep changing their naming system). Anyway, it has so much more capability than the original, that the CRS-10 missions are not very heavy (also a huge factor is that the second stage only has to get it into Low Earth Orbit, and not a Geosynchronous Transfer Orbit).
What I think they did for this one, was adjust the flight path so that the first stage would climb at a higher angle (loft) than it would for a mission without an RTLS landing. So, its parabolic (ballistic) path after staging would not be as far downrange, so it would not need to use as much fuel to kill that downrange momentum and boost back. Now, that would mean then that the 2nd stage was not traveling as fast horizontally as usual. But since the Dragon payload is not that heavy, and it "only" needed to get into LEO, it was able to get into LEO despite a slower staging velocity. And, to account for Earth Gravity trying to make it fall down (or not climb as well) more due to the slower staging velocity, the greater upwards momentum from the higher angle (loft) solves that.
I experienced that sort of thing in Kerbal Space Program. Ultimately building a reusable booster that just flew vertically and staged with a lot of upwards vertical velocity, zero horizontal velocity, as the second stage slowly accelerated horizontally after staging. The second stage would fall down into the atmosphere if not for the high upward momentum of the first stage. Of course that was an extreme case, not necessarily practical. But as often happens, a great way of learning some aspects of rocket science and unforgiving orbital mechanics.
A better real-world example was the Space Shuttle. After SRB sep, the Orbiter and ET had a relatively low thrust to weight ratio. So, the SRB's accelerated the shuttle to a relatively high loft, after SRB sep the upward momentum gave the orbiter and ET time to accelerate horizontally to near orbital velocity before falling back into the atmosphere.
So, that would be a contributing issue for the booster landing sooner than for the at-sea landings. Of course, it really does not matter in the big picture if the booster lands at mission elapsed time 8:08, or 9:30, or a half hour later. As long as it lands safely where it's supposed to.
BTW - The original Technical Webcast on youtube, it ended after 15 minutes, before launch.
SpaceX reposted the Technical Webcast, the corrected link is below. It is worth re-watching the flight. They did have an at-landing-pad side view video of the landing, on the right. On the left, the onboard video, which had a lag time of a few seconds as on the right the ground camera shows it landing, while on the left the legs are not deployed yet. I had noticed the lag in the Hosted webcast by the crowd reaction to events like staging, could hear them react before anything happened with the onboard video:
[video=youtube;rUDLxFUMC9c]https://www.youtube.com/watch?v=rUDLxFUMC9c[/video]
Below, a photo sequence of the landing , posted by Mseely1 on Reddit:
https://www.reddit.com/r/spacex/comments/5v29n7/crs10_landing_seen_from_the_nasa_causeway_45/