Add rule to resolve potential rematch scenario in round 3

[claabs]

With the introduction of the new Initial Swiss Matchups rule added for the 2024 Copenhagen Major cycle, a flaw with assigning round 3 matchups was introduced.

There's a 0.6% chance that 1-1 record teams in round 3 will force a rematch, which is not allowed by the current ruleset.

Example Matchups

Here's an example round 3 1-1 matchup pool. First we seed based on the mid-stage seed calculation rules;

After the initial matches, seed is determined (in order) by:

1. Current W-L record in the stage
2. Difficulty Score in the current stage
3. Initial seeding of the current stage

Team	Difficulty	Seed	Previous Opponents
1	2	1	1>9, 16>1
5	2	5	5>13, 12>5
7	2	7	7>15, 10>7
11	2	11	11>3, 6>11
3	-2	3	11>3, 3>14
9	-2	9	1>9, 9>8
13	-2	13	5>13, 13>4
15	-2	15	7>15, 15>2

Then we follow the rule for round 3 matchup assignment:

In round 2 and 3, the highest seeded team faces the lowest seeded team available that does not result in a rematch within the stage.

The matchups are created from highest seed first, picking the lowest seed that doesn't result in a rematch:

• 1 is top seed, so they pair with the bottom seed 15 first
• 5 is next highest seed, but already played 13, so they pair with the next lowest seed, which is 9
• 7 is next highest seed, and pair with the lowest seed available, which is 13
• 11 and 3 remain, but they already played each other in round 1.

Resulting flawed matchups

Team A	Team B
1	15
5	9
7	13
11	3

This conflicting matchup is not present when using the old seeding rules (1v16, 2v15, etc). There is currently no explanation in the rules how to resolve this scenario.

Solution

To fix this, the rules should be updated to either:

1. Add a clarification to resolve the scenario where rematch conflicts appear (e.g. swap opponents for the last two matchups when a conflict occurs)
2. Revert the Initial Swiss Matchups rule
3. Seed the teams using the Initial Swiss Matchups rule throughout every round of the Swiss event

I chose the first option in this PR, as to preserve the Initial Swiss Matchups rule. There are other ways to resolve the rematch scenarios with varying factors of "fairness". This is the simplest.

Background

I wrote a Monte Carlo simulator to simulate every potential event outcome, which is how I found the flaw. The source can be found here: https://github.com/claabs/swiss-buchholz-problem