The bitstream with any plain (ID-unaware) drive connected would be identical to no drive connected at all, that is all 1's, for the very reason you just mentioned.
RDY is active low - so the CIA will keep it high (=1) until something connected to it signals by driving it low (=0). So both no drive at all AND a connected but detection-unaware drive would, as you correctly state, leave RDY untouched, resulting in a stream of 1's.
An easier argument: If it was as you say, then the separate DD-ID circuitry would not be required for external DD drives, nor would e.g. (on the A2000) the J301 jumper be needed to enable detection of a DD drive as df1:
If you take a look at the A2000 schematics, you can see that all J301 does is bridge the ID circuitry to the RDY line.
With J301 closed, the RDY is fed: ~(~SEL_1 & Q) = SEL_1 || ~Q
Now before the ID detection routine enters the bitstream-reading loop, MTR is driven high and SEL_1 low, then in the main loop before each reading of RDY a high to low transition of SEL_1 is executed, resulting in a low to high transition on U108's clock input. This latches the state of MTR into U108's D input, and Q = D, therefore Q becomes MTR each time the loop is executed and since MTR is not changed throughout the loop it remains fixed to its initial high state.
Thus, right before RDY is read, Q = high = 1 and SEL_1 = low = 0, and thus RDY = SEL_1 || ~Q = 0 || 0 = 0 and therefore we get a stream of 0's as we indeed should for a 3.5" DD drive.
But the same procedure for a plain drive as unit 0 will yield a stream of 1's. A way to resolve this is to make an assumption on the drive type, the natural one being it's a 3.5" DD one as all pre-HD Amigas came with one installed anyway.
