>[A]re there still many industries (or just individual tasks even) that are necessarily locked into proprietary products because of complexity/cost of development, lack of interest outside of a niche of industry, or whatever other reason.
I would say almost all paid software development occurs within these niches, actually.
NOTE: I need to slightly redefine your question as "Are there any good open-source options", because in some some sense all software has a Minimum Viable FOSS Competitor: Build it yourself from scratch.
But one very concrete example might be MATLAB. You may assume that GNU Octave, Julia, etc. are all good competitors to MATLAB, but the real value add of MATLAB is that the company behind it has funded high-quality, proprietary libraries for all kinds of niche scientific and engineering toolkits. It is much more difficult to whip up the equivalent FOSS enthusiasm to put the same TLC into a sub-sub-category of optimizing a control theory equation, and that's if you can find a FOSS dev who even has the math background to understand what's going on there.
Two other areas which come to mind are ERPs and capital-R Requirements Management software. Both are relatively niche themselves, with hundreds of individual rabbit holes to go down. If there's a unifying thread, I would say it's this: Truly valuable software combines good-enough software engineering with good-enough domain expertise. If all you ever do with your time is hack, you only know the hacking part of the equation, and that's usually not enough to be sold well by itself.
