It's not a constitutional convention unless relevant political actors and authorities are in general agreement about the rule.
Precedents alone do not make a constitutional convention. You also need general agreement about what the rule is and that the rule is binding. You also need a reason for the rule. Typically that rule is tied to a higher constitutional principle.
When it comes to the 'most seats = PM', we have precedents. We also have contrary precedents, even recent ones, though they're admittedly at the provincial level. Do we have agreement? Not really. Authorities certainly don't agree and not all political actors do.
What about reasons? We might try to tie it to notions of popular democracy and fair play. But we can't tie it to the higher constitutional principle of parliamentary democracy. The confidence convention, on the other hand, is tied to that principle.
So, the 'most seats = PM' has precedents. It also has agreement in some circumstances, but not all. It's not evidently binding either. And it has difficulty hooking into a constitutional principle that gives it a reason.
So what is the 'most seats = PM'? I've argued that it's a custom. That's a kind of rule, but not one that's constitutionally binding. It's how we 'typically do things' but not always. There's a basic agreement that we should follow it most of the time, but it can be set aside.
Here's how I define custom in a recent article and how they differ from conventions. I also discuss the particular custom of the party with the most seats governing and why it isn't constitutionally binding, though we usually follow it. 
By way of contrast, let's consider the confidence convention. I would argue that most political actors and authorities agree that if a government loses a vote of confidence, the PM must either resign or request a dissolution.
The purpose of resignation or dissolution are to either a) find another government that can command confidence or b) to elect a new House of Commons that may express confidence in the existing government or a new government. The goal is finding a govt that has confidence.
We have basically all the precedents we need here. Harper in 2011 is a recent one at the federal level for a vote of no confidence followed by a dissolution. Clark in BC is a precedent of a vote of no confidence, a request to dissolve denied, resignation and a new govt.
I don't think there are political actors out there who would argue that a PM could stay on if they were defeated on a confidence vote, got a dissolution, and lost yet another vote of confidence in a new Parliament within 6 months and there was a viable alternative govt available.
So, we have the precedents and the agreement. What about the reason? The reason for the confidence rule is connected with the constitutional principle of parliamentary democracy.
Since our governments arent directly elected, our parliamentary democracy rests on the executive being able to demonstrate that it is supported by the elected house of Parliament.
Here's how I describe the confidence convention in the article.
And to complete the argument, here's how I define conventions.
And how I define customs.
Finally, I adapted my tables from the article to deal specifically with this question: lagassep.com/2019/10/17/who…
Oh, and I use the same criteria to explain why the PM’s right to stay and test confidence first, despite the election results, is NOT a convention. It’s a legal reality supported by other conventions. We could arguably change that legal reality with a law: lagassep.com/2018/09/28/sti…
Specifically, we might be able to pass something like the Government Transition Act, either as a regular statute or a section 44 amendment, that would require the PM and Cabinet to resign following an election.
Following their resignation, that same PM could be commissioned to form a new government by the GG, or the GG could turn to an alternative PM. I think we could do this without triggering s41a of the amending formula.
