Good to hear your new Candy is indeed much better. There are good 5ch amps out there but they cost £1-2k upwards - hardly surprising when you have 5chs instead of just 2 & even more careful design of the power supplies being needed.
Bi-wiring is one of the subjects, along with 24bit/96-192k coding, expensive interconnects, cable lifters etc. etc that have either no or little foundation in physics. I'm beginning to wonder if some "improvements" are so small in everything but cost that they are effectively worthless - even when real rather than imagined. What I'm suggesting is that there's a threshold of improvement below which other aspects of life just mask any effects such as mood, tiredness etc.
If a mischievous friend placed shorting links across my bi-wired terminals, how many months would it take for me to realise? Quite possibly not until I wiped dust off my speakers!
The issue with small improvements whether real or imagined is that they tend to disappear with blind ABX testing. The problem with audio is memory - you can't really compare A with B by feeding A to the left & B to the right channels simultaneously but must switch both in time. Comparing 2 pictures side by side is much easier, for instance. Auditory memory is notoriously unrealiable. Some may maintain that ABX testing is itself unreliable but it gets to a point where people just don't want to let go of cherished beliefs - they effectively turn their hobby into a religion.
As for speaker cable thickness, this is another disputed area. Some maintain that as long as the resistance & inductance are kept under certain maximum levels so as not to effect the frequency response - the thinner the better.