OK, so I did a bit more groundwork to investigate what we can do.
First question: The max temp drop above was more or less 98C -> 90C or about 8C. Maybe that shot was a little short; let's give ourselves a margin of error and say it's 9C. But the max temp drop is limited by (1) the minimum 80C temp set, which influences the group catridge heater as well as the water temp and (2) how much heat the system can actually lose; its insulation. So we speculated that the max temperature gain that is possible might be able to exceed the max temperature loss. That's all background. Is it true? Answer: Yes:
We can go from 80C to 91C, for a temp gain of about 11C. This was over a 50g shot.
Next question: Having established that we can add more temp than we can drop, is creating mirror image temp profiles as simple as setting the starting temp of the increasing temp profile to the end temp of the decreasing temp profile and vice-versa for the ending temp? Answer: Yes:
In the above example, I cut the shot ramping up at about 43g and, consequently, it didn't ramp up quite high enough to more or less reach the starting temp of the shot ramping down. Had I not preempted it and let it run for about 50g, I think we would have gotten a slightly more satisfying graph.
For completeness, I should point out that I dialled in the sacrificial coffee for this exercise to produce pressure profiles that look like what I like on the blooming espresso profile using my grinder, so, in that sense, the conditions are realistic:
Aside: The second pressure peak is what I use as my secondary reference point when dialling in (the primary reference point being what's in the cup, of course). The slope of the pressure dropoff from that peak seems to be a characteristic of the grinder/burrs (well, assuming that there hasn't been some sort of barista error in pulling the shot).
Bonus round: Having this pressure profile programmed, I wondered what would happen if I simply deleted the "bloom"/soak phase (ie. after preinfusion when the pump isn't running). The answer is that the pressure jumped up a lot, dropped off quickly, and a large part of the shot volume was pulled at a very low pressure (and looked terrible). Dosing 4g less wasn't enough to lower the pressure peak significantly, nor was grinding a whole number coarser on my EK ... on my particular dial (sigh ... can someone please make a 360 degree EK dial that becomes the new standard, so that we can talk in objective, consistent, translateable numbers?) and neither remedied the marked transition between a very high pressure phase and a very low pressure phase. I think Rao has said that the bloom allows the use of a finer grind setting and this exercise certainly showed that. So I don't think that this particular profile is going to be a productive thing to investigate. Anyway, here's an ugly graph for you of these ugly shots. Parental guidance recommended:
I feel all of that gets enough background out of the way that I can now make some shots, taste them and report back. We've also established that for these sorts of temp profiles, we only need two temp segments, so later on I can probably program a pressure profile that's more generic, like a standard e61 profile or something, with an overlaid temp profile. That way, we can sort of take more oddball temp/pressure profile out of the equation. Actually, if I can do that easily, I might do that before tasting anything and use it for the shots that I taste.
... and after all of this, I'm seriously considering referring to the machine as the temp-a-sketch ...