Some early efforts were made using print film and real scenery, but easier to interpret comparisons can be made using a specially made target photographed at a distance of 40x the focal length on test.
The above is a sample of the test pattern I made using the laser printer. It was created pixel by pixel in Paint Shop Pro then then repeated to make long strips and the canvas size altered until it printed out with the smallest line pairs yielding 1 line pair per millimetre, long strips were made and then glued diagonally across a grey painted board 1440x960mm. When photographed at a distance of film plane to test pattern of 41x then the pattern is reduced 40x on the film. Then direct examination of the film yields 40lp/mm, 20lp/mm and 10lp/mm performance, with the 10lp/mm being the largest marks of course.
The pic above is of the whole test chart, 1440x960mm (which is 40x the standard 35mm frame of 36x24mm). The white tape at the edges helps evaluate distortion, the circles correspond to 5,10,15,20mm from the centre. The four white dots near the centre are Velcro spots and are used to hold a mirror in place to get true alignment to the camera. The three Velcro dots on the bottom left are used to hold cards saying lens type, focal length and aperture so there is no need to write notes and get mixed up later. The rough red mark is the area shown below.
Above is a 4,000 dpi scan and compressed by JPEG by 15% with no sharpening at all of what you get to see when evaluating the slides of the test chart. This bit is on the 5mm line as shown in the previous pic. The film used is Fuji Sensia 100, so the test represents a realistic use of lens and film.
My notes are not with me but I know it is the Cosina 100/3.5 macro lens at f/8. To use up the end of a film I didn't bother with making the test cards to show what lens/aperture it was. To me that is reasonably good resolution by comparison with some other lenses. The inner 40lp/mm bits are not too degraded.
For comparison, above is the same area from the same lens at f/3.5, maximum aperture. The whole slide shows a large falloff in light from centre to corner and the sample here shows a loss of clarity due to flare and loss of contrast that clears up as the lens is stopped down.
To show the worst possible performance of the lens, above is the very bottom left corner of the image at 20mm from the centre with the Cosina 100mm lens at f/3.5. Clarity is really getting bad now, but it is still a lot better than some lenses I've seen. The crinkle on the bottom is in the white plastic tape that is coming unstuck in the warm sun.
Some macro lenses were also tested at 1:1 ratio by using a borrowed microscope resolution test plate.
Meanwhile the only test/evaluations are the old ones that have been around for while on this site.
Nikon 24-120/3.5~5.6D Vs Nikon 28-105/3.5~4.5D
Nikon AF 75-240mm f/4.5~5.6D
Nikon AF 28-105/3.5~4.5D - A Macro Performance Evaluation
Tamron teleconverter test. Newly added
In general the comment can be made that most lenses always perform better around f/8 to f/16. But you do need to test at all apertures and at all focal lengths to prove that. From about f/16 to f/32 or whatever the minimum aperture is, the image gets worse due to diffraction effects, just a fact of life with all lenses. The very worst quality usually comes when you use the lens wide open as then you are including all the aberrations of the whole lens surface and also the very shallow depth of field may also yield an effect that is not what you want.
At 1:1 macro distances it is very obvious how the image starts to degrade once the aperture gets smaller than f/8. Diffraction seems to hurt macro more than normal work.
One of my recent tests (unpublished here yet except for a few bits of the Cosina test above) was to check my Nikon 105/2.8 macro lens against the very cheap Cosina/Vivitar/Soligor/Phoenix 100/3.5 macro lens. The results at maximum aperture of the cheap lens were bad, but once stopped down to f/8 the results are very nearly as good as the very expensive Nikon. The Cosina is great value for money and it is much smaller and lighter. Also its working distance is longer as it doesn't use internal focussing like the Nikon. It seems the actual focal length of the Nikon changes from 105mm at infinity down to 79mm when focussed at 1:1 due to its internal focus operation.
This focal length shift is also very evident in super-zoom lenses where the indicated focal length is all rubbish when focussed at close to medium distances. I suspect the Sigma 28-300 lens is actually a 28-200 lens when focussed at its minimum distance. Proper tests are yet to be done to prove that.