How was 8080->8086 translated code developed further?

When IBM released the PC, many programs, such as WordStar if I recall correctly, were ported from CP/M to MS-DOS using 8080 to 8086 translation tools like the ones discussed here:

Once translated, how were these programs maintained and further enhanced?

Were the binaries just disassembled, kept in Assembly, and edited to fix bugs and add new features? Did the code understanding gained from the translation feed into an eventual rewrite in higher level languages?

What do you mean, “disassembled”? There was no disassembly involved. Translation by the Stage2 macro processor was at the assembly language source code level, not at the binary level. The products then in 8086 assembly language were maintained at the 8086 source code level for a short time as long as there was a market for them. But the limitation of the 8080 64k memory model meant the product was not longer attractive due to a market where the 8086 memory models meant that far larger projects could be built. The Fortran product was replaced by one written in a higher level language that made full use of 8086 memory model capabilities. At that time (1982) there was no one at Microsoft willing to rewrite the 8080 model linker to expand its capabilities.

Thanks, I wasn’t sure whether the conversion tools worked at the source or binary level. So I assumed they were fed 8080 binaries as input and produced 8086 binaries, which may have been simpler.