Using the 1 Hz signal as the trigger will indeed take a long time as it will only be acquiring one sample per second.
The scope generates a random sample delay from the trigger to sample what it sees on its sample head inputs. I think it won't update the screen until every slot has been filled, but I'm not sure about this.
There is a minimum delay of 25 ns from the trigger to when it samples, so you would not be able to see the edge of the 1 Hz signal if you were to simply split it into both the trigger and a sample head input. Of course, you would need a proper 50 Ohm splitter, not just a Tee.
Tektronix sold delay lines for this purpose that included both the splitter and the delay required to see the trigger event with a scope channel.
https://w140.com/tekwiki/wiki/DL-11The 11802 model had the delay lines built in.
You need to make sure that your trigger settings are set properly. It's a bit finicky. I would suggest you start with a faster trigger signal from a function generator to make sure your setup is OK.
In addition, you will run into the problem that a 2.5 MHz signal is still very slow (400 ns period) and unless the 1Hz edges are expected to be very close to the 2.5 MHz edges you would have to use fairly slow horizontal timing settings since the maximum samples of this scope in a trace is around 5000. This would kind of defeat the whole purpose of using this type of instrument.
I would suggest to just try to measure the jitter as the trigger to 2.5 MHz delay, rather than trying to catch both 1 Hz and 2.5 MHz edges in the acquisition.
You first zoom out to capture >400 ns and then adjust the delay and the time base to zoom in on the edge transition.
It might feel a bit sketch as what you are measuring is the sum of the signal jitter, trigger jitter and delay jitter. However the latter two are specified in the 11801C data sheet, but I'm not sure if this was guaranteed down to a 1 Hz trigger rate.