I came up with almost the same idea as the editorial . The only difference is that I queried $$$i$$$ and $$$i+1,i+2,...,n$$$ to find a magnet (I think it will be either the first or the second of them),instead of $$$i$$$ and $$$1,2,3,...,i-1$$$.And then I used the same approach just as the editorial says.

However,I got TLE on test 6. qwq

Today,I did a binary search on test 6.I get the test case that made me TLE, but it passed when I tested it locally.

I do wonder why it goes wrong.

You check my submission here 108735158