In problem C, apparently very dumb tests before hacking stage. My solve 359774235 pass tests but crush on a lot of tests, for example if a[i]=1 and b[i]=0 for all i in [1,1e5] or a[i]=1e5-i and l[i]=1,r[i]=n my solve do O(n^2) steps. Tests selected partly by peoples or by LLM or random? And its the rare case or need to be more careful then I write solves on easy problems?