This is a night and day difference with respect to reverse engineering...
No, it isn't. They had to go further out of their way to dance around that issue in order to make a legal clone.
The half of the clean room effort that does the implementation are the one's making the clone, they don't see source code, disassemblies, etc. The other half doing the reverse engineering in order to develop the specification have to discover the *intent* of the original developers with respect to functionality. That discover is easier when you have their commented source code rather than a disassembly of a binary.
The dancing you refer to is for non-clean room scenarios where the developer implementing the compatible non-infringing clone has access to the original copyrighted code. And that dance occurs regardless of whether he/she is working from a binary disassembly or commented source code. Lawyers literally look at the code and say these ten or so lines in the new are too similar to these ten or so lines in the original. Disassembly or source has this same problem. Now source still has the advantage of better divining the original intent, so having the source is also a win in the non-clean room scenario.
...and the fact that IBM didn't want a compatible BIOS to be produced does not change this.
It changes this part:
Compaq et al were able to create clones because the IBM PC was an open platform.
No, it didn't. The fact that IBM provided source code to all PC programmers as a way of documenting the BIOS API actually made things simpler despite such a desire. If IBM was to act in a manner more consistent with that desire so as to hamper Compaq et al they would have simply provided PC programmers with registers for input/output parameters and the interrupts to use to invoke an API call. As was done with DOS.