In eukaryotic transcription will the nucleosomes ever completely unwind the DNA and the histone complex disassemble? If an operon is more 160 base pairs it seems it must.
Yes, nucleosomes are completely unwound. Histone chaperones such as FACT (for H2A/H2B) and ASF1, CAF-1, HIRA, Nucleophosmin etc (for H3/H4), associate with RNA Pol II and handle the displaced nucleosomes. As you surmised, the histone octamer complex is disassembled, into the H3/H4 tetramer and two H2A/H2B dimers. Right behind the elongating Pol II, the chaperones reassemble nucleosomes. During the process of reassembly, it has been shown that "new" H2A/H2B dimers (and possibly H3/H4 tetramers too) can get incorporated into an "old" histone octamer. Finally, nucleosome remodelling complexes such as SWI/SNF re-space the nucleosomes.
This is a decent review of the process (by one of the discoverers of the FACT complex, incidentally): "New chaps in the histone chaperone arena". Campos & Reinberg. Genes & Development (2010): 24: 1334-1338
This is a comprehensive review of histone chaperones and nucleosome dynamics during various DNA-dependent processes (such as transcription, replication and repair). There is an entire section on their role in gene transcription, which is the main thrust of this question: "Histone Chaperones: Assisting Histone Traffic and Nucleosome Dynamics". Gurard-Levin, Z.A., Quivy, J., and Almouzni, G. Annual Review of Biochemistry (2014): 83: 487-517
And finally, a good overview on histone dynamics and chromatin structure during transcription: "Histone exchange, chromatin structure and the regulation of transcription". Venkatesh & Workman. Nature Reviews Mol. Cell. Biol. (2015): 16:178–189
I should add that an inability to disassemble nucleosomes (or displace any other factor bound very tightly to DNA) can actually pause an elongating Pol II complex.
Just to add very recent new information to the first answer: the initial steps of RNA Pol II transcription through a nucleosome have now been revealed at the structural level by cryoEM.
See these two articles: