Isn't that scheme fallacious?
Your "ridiculously long" p/w is no stronger than the weaker, albeit "much stronger" one that unlocks your high-security keychain.
My login p/w is ridiculously weak, but there's no risk involved, because my deuced Mac(hina) is a one-person machine.
For peace-of-mind, rather than immediately necessary security, though, I use a much stronger p/w to unlock my keychain and a very significantly stronger one to unlock my sparse image.
It would be fallacious only if I thought the ridiculously long password was the one protecting the disk image. I know the password of the keychain is the weak link, but I chose it to be strong enough for my needs. When I created the disk image, I let the system auto-generate a password what it believed to be a maximally secure. But such passwords are actually weaker; any password you cannot remember must be written down somewhere, and is only as secure as wherever it's written. The solution to this conundrum is to write it somewhere it's protected by a password you can remember.
This is the way password managers like LastPass or 1Password are usually used. The user lets the system auto-generate unique passwords that the user never intends to ever enter or even remember. It's up to the password manager to remember them, and the strength relies on a single "good enough" password that protects the password manager's data. Using a single password to protect all the "real" passwords makes that password easier to remember, by dint of being entered more often. You get the convenience of only having to remember one password with the security of having separate passwords for separate uses.
I'm too cheap to spring for a commercial password manager. This is how I set up my own system along the same model.
