I would be interested in getting the response from an objective manufacturer. Here is my take, which is slightly different than @Hazen_Babcock.
Higher NA means higher collection angle by definition. The amount of light collected from an isotropic emitter is proportional to NA^2 (e.g. from an ensemble of randomly oriented fluorophores in fluorescence).
Recall that NA = RI*sin(α), so NA < medium refractive index… unless you figure out how to collect light with more than 90° half-angle. The Olympus 1.7 NA uses a special coverslip and special immersion oil with RI > 1.7. The medium with refractive index called out in this equation is the medium the objective is designed for (i.e. the immersion medium for oil objectives).
Objective lenses that have NA/RI ratio anywhere near the maximum of 1.0 (i.e. large collection angles) usually have very short working distance because otherwise they would have a huge front aperture. TIRF lenses fall into that category. I believe than a NA 1.49 TIRF (standard oil) objective really does have a arcsin(1.49/1.5128) ~= 80 degree collection angle, but it has a tiny working distance (BTW the working distance specified is the free working distance besides the coverslip, if one is specified). For TIRF you are imaging right at the coverslip surface anyway so you don’t need more than the 10s of micron WD anyway.
Objective lenses also have transmission curves which tell you how much light gets through. As I understand the main loss mechanism is residual reflections of the AR coatings on internal lens elements (which might be e.g. 1%/surface). It is conceivable that an objective with lower NA but higher transmission might win the competition for collecting the most photons from an identical sample.
To your question of whether having more NA will compromise image quality, I would say “no” but also be aware that objectives have been designed to perform different tasks and use caution when trying to repurpose them. For example, I am aware of several objectives designed for multi-photon imaging where the PSF size on a camera is significantly larger than the NA would predict. For multi-photon imaging this is acceptable, but if you use on a lightsheet system beware. So indeed it is possible for lenses to have intrinsic “underperforming” resolving power relative to the NA-based theoretical limit. I am not aware of any TIRF lenses with this problem. It is also possible for objectives to get damaged or poorly assembled in a way that they don’t achieve anywhere near the theoretical resolving performance.
Different objective lenses have different corrections depending on the target application. Spherical aberrations are the most common type of imaging aberration and occur as you image deeper into a sample that isn’t index-matched with the immersion medium (hence the popularity of silicone oil objectives where the immersion substance better matches the RI of many samples than either water or conventional oil does). Some objective lenses – depending on intended use case and sensitivity to spherical aberration – have correction collars so the user can adjust internal compensation. Some objective lenses have better flat field (plan) correction than others. Some objectives have better chromatic corrections than others (achromat = corrected for 2 colors, fluor = 3 colors, apo = 4 colors). TIRF objectives are usually well-corrected for flatness and chromatically, but multi-photon objectives often do not perform particularly well in either of these categories…
As a final note, I would say to a certain extent you have to try it out yourself to see if a particular objective works in your application. As Hazen said, sometimes you can arrange to test out objectives.