Yeah, I know what you mean. I usually leave out the gasket when tuning the Flexiwhip. The Larsen NMO-Q is the same way, with a fixed center conductor. So I want to make sure the antenna is making solid contact with the base while tuning so I know it's good. Then I recheck after putting in the gasket and smearing some dielectric grease to make sure nothing has changed.
The difference between good and bad with that gasket is very slight. The generic rubber gaskets you find in hardware stores are usually too thick and don't compress enough, so be sure to use the one they include or one just like it. I also found that the hardware store plumbing gaskets deteriorate and turn to goo after a couple of years being exposed to UV and mag chloride...
I may have mentioned, but you have to interpret the meter readings. A 50Ω resistor has a prefect load match but radiates almost zero RF. So you may not have or even want to see a 1.0:1 SWR reading. My MFJ-269 only has I'm guessing a 5% precision, so 1.0:1~1.05:1 is what I read into a calibration load and precision gets worse at greater than 2:1 mismatches.
That 2.2:1 or so SWR does not surprise me, the bandwidth of a 3/4λ antenna is narrow, so if you've dialed it in for a 2m repeater on the high side it's quite likely you'll be on the long side of the curve for 70cm repeater inputs. See posts #21 and #24, when SWR at 147Mhz is 1.4:1 the same antenna is 2:1 at 440MHz. Those NEC models don't factor in parasitics or discontinuities inherent in a real antenna. Those are just feedpoint characteristics. We already know the UHF connector is lousy at UHF frequencies (remember 'UHF' when the PL259/SO239 was designed in WWII was what we now call VHF), it's likely the RG58 is start to get lossy at 440MHz, etc.
VK3JEG ran some tests on the UHF connector in fact: http://www.qsl.net/vk3jeg/pl259tst.html
I was suggesting you add a short length of wire, like 1/2", between the coax and antenna. That's not a practical solution, though. It's very common with mobile HF antennas to put coils at the base to change impedance because being physically short relative to wavelength yields extremely narrow bandwidths. So you'll be dead on in the phone spectrum but completely mismatched in the CW and data. So you add reactance to bring resonance down or up. You might also be able to add capacitance but it's easier to add inductance. I'll have to think about it, it might be possible to add a coil at the top end or maybe a parallel cap. Problem is mechanically how to make it work.