Can't remember if the M1 has true dialtone detection off the top of my head,I don't think so, but basically speaking, a dial attempt is any time the panel goes off hook, dials and doesn't get a handshake or kissoff, so sometimes even if there is a functioning line, if the message doesn't get through, it'll hang up and try again.
I haven't had to listen with a butt set on the M1, but on others usually there's some pausing and downtime, not to mention the 10-11 digits for most standard CS numbers to dial using DTMF. In some cases, the dialing, waiting and all that for 8 attempts can be 15 minutes or so.
In the specific of a burg panel, it'll dial or attempt on the primary until there's no attempts left or if the panel knows there's no line there (line cut monitor) it won't even try the primary.
I can't remember if the M1 will look for a line if no POTS dialer is programmed or not or if it's a function of the RC for telco line fault....can't say I've ever had a panel not connected to POTS or VOIP/Digital voice as a backup to a Uplink....and I use those over dialer capture units on the M1 because of the bidirectional supervision, though it means a more difficult time in latching a trouble condition on the M1 (relay to zone) so the end user is made aware of an issue with the Uplink or serial on the M1 keypad.
In your case, with the VOIP, I might keep that at a minimum and then cell, but I'd definately change the sequence unless you can prove the voltage on the ATA goes away when no network connection is present. It still doesn't address any latency, but it's still not a best practice.
And thinking back to some heated discussions regarding the HAI C3...that is really a worrisome situation if you have the C3 in line with your phone service and it's VOIP.....the C3 is never going to be switched "on" and used if the voltage is there from the ATA and there's no true dialtone detection on the panel. Just a thought to those out there, even though I did provide a host of reasons why I don't like the C3 as a secure option.