Unlike English, it's possible to read Czech without knowing the words (no tricks like "tomb"). If you omit accents, one would either have to know correct pronunciation or he will read it differently.
What you could do is to replace an accent with an ASCII character. Compare the name of town Český Těšín with its Polish equivalent Czeski Cieszyn (Č → Cz, ě → ie, š → sz). Actually, this is how we got here, 600 years ago Jan Hus (John the Goose) added diacritics to shorten a few diagrams in written language.
The difference between `s` and `š` is similar to `s` and `sh`. Can you omit `h` in Englis? Well, tecnically yes, but you canged the language. Same on you!
Yes. I'm also using my own layout on ErgoDox EZ, my languages are Czech (příšerně žlouťoučký kůň úpěl ďábelské ódy) and English. I'm also a (neo)vim user.
I found it hard to switch layouts, I'm using only one with layers. On first layer there are english letters, second layer there are accents (a → á) and some symbols where accent is missing (m → +). On third layer there are arrows on home row and functional keys. Shift without other key is `(`, on second layer `{`. There are some additional keys at the bottom and middle, they cover the rest of symbols. Outliers are mapped to start editor, browser, ctrl+c, ctrl+z, volume, mute, …
I really like having a single layout. I don't like that I'm no longer able to type on QWERTY without looking at keys. On my desktop, I'm 10 % faster, on any other computer I'm 90 % slower. Would I do it again? Probably yes.