fix: cherry picker weighted latency formula

[crisog]

This PR brings in fixes suggested by discord user msa6867#1823.

Given the user's comment about not being comfortable with Git, @blockjoe wanted to just pull in the changes since they were quick/simple so that the user could say if this is what they were suggesting.

Issue:
Nodes past 20 relays mark (majority of them) are being calculated at 100% median latency + 30% P90 latency.

Actual Formula (100% median + 30% p90)
Name	Median	P90	Weighted Latency
Node1 - 100	135	100 + 40.5 = 140.5
Node2 - 112	119	112 + 35.7 = 147.7
Node3 - 115	117	115 + 35.1 = 150.1
Node4 - 125	126	125 + 37.8 = 162.8
Node5 - 120	123	120 + 36.9 = 156.9

Proposed Formula (70% median + 30% p90)
Name	Median	P90	Weighted Latency
Node1 - 100	135	70 + 40.5 = 110.5
Node2 - 112	119	78.4 + 35.7 = 114.1
Node3 - 115	117	80.5 + 35.1 = 115.6
Node4 - 125	126	87.5 + 37.8 = 125.3
Node5 - 120	123	84 + 36.9 = 120.9

Note: this is artificial data, not from an actual session.

As you see, using the actual formula you take nodes out of the top bucket with both median/p90 within requirements (<150ms). After the adjustment, this will no longer happen.

[height]

Link Height tasks by mentioning a task ID in the pull request title or commit messages, or description and comments with the keyword link (e.g. "Link T-123").

💡Tip: You can also use "Close T-X" to automatically close a task when the pull request is merged.

[msa6867]

Second Issue:
weightedSuccessLatency is used for both sorting and for calculating weighting; on the other hand, the main role of medianSuccessLatency is merely as a stepping stone to calculate weightedSuccessLatency. Thus it is logical inconsistency to use medianSuccessLatency rather than weightedSuccessLatency in a conditional statement to calculate the weighting. This leads to a node whose median is equal or under 150ms but whose weighted is over 150ms to avoid weighting penalization. For example:

Actual assigned weight assuming first issue is fixed (latencyDifference conditional on medianSuccessLatency)
Name Median P90 Weighted Latency latencyDifference WeightFactor
Node1 - 100 135 110.5 0 10
Node2 - 150 650 300 0 10

Actual assigned weight assuming first issue is fixed (latencyDifference conditional on weightedSuccessLatency)
Name Median P90 Weighted Latency latencyDifference WeightFactor
Node1 - 100 135 110.5 0 10
Node2 - 150 650 300 150 5

Note: this is artificial data, not from an actual session.

As you see, after the adjustment, node2's weighting is properly reduced