Shouldn't profit be proportional to liquidated amount?

[Shymaa-Arafat]

I'm trying to get the most learning value of this exercise and have tried different options:
-In systematic 2 steps full liquidation (liquidating 1568 ETH), I get profit of 24.11 ETH
https://github.com/DrShymaa2022/defi-mooc-lab2
-If I compared this to 1 step profit as 21...ETH, it seems reasonable and satisfiable to increase it by about 1/7 = 14%
-However, if we looked at the details of the results 2 annoying observations:
1-How come a gain of 142 ETH ends to be only 24.1?
2-How come what a lot of people did of liquidating less than ≤1000 leads to more than double the profit (43.8 ETH, for difference gain ranging from 86 to 94 if we applied 2 steps
https://github.com/Shymaa-Arafat/defi-mooc-lab2/blob/a00d99b0dd63c1da3cf32d9b918f63b8bfcd72c0/contracts/LiquidationOperator.sol)?
.
-If the flash swap charges ratio "rt" for the Flashloan, ie for loan X we payback (1+rt)X { rt<1 in fact should be 0.003?}
If X1>> X2
X1 leads to 142 ETH
X2 leads to 86 ETH
Profit Pr= gain from liquidation - ratioX
for Pr1 < Pr2 ( what happened in the runs)
rt*(X1-X2)> gain difference
in our case
rt1.5m$ >142-86 ETH
rt> 561.8/1500
rt> 0.0672
no way uniswap charges more than 6.7%????
If we added more exact values ( difference is nearly 20 ETH)
It should be rt = 76*1.8/1500 = 9.1%
.
If this is not the case as I think, please take the time to find out what is the common mistake most of us had fallen in and was not discovered by the test script
( I tried running other people codes and changing parameters or printing details to check it out) and was not discovered by the test script
.
-Finally, would be helpful for the next course version if the read me guide shed some light on using uniswap router or regular call to payback and pros & cons of both ( people in discord mentioned re-entrancy attack)
.
Thanks