Image comparison test failures should explain how to investigate them

[kpreid]

If an image comparison test fails, the error text is like:

        FAIL [   4.647s]                     wgpu-msaa-line-example::bin/msaa-line msaa_line

--- STDOUT:                                  wgpu-msaa-line-example::bin/msaa-line msaa_line ---

running 1 test
Starting image comparison test with reference image "/Users/kpreid/Projects/rust/contributing/wgpu/examples/common/../..//examples/msaa-line/screenshot.png"
Mean: 0.075663
Min Value: 0.000000
      25%: 0.219296
      50%: 0.371955
      75%: 0.475240
      95%: 0.623761
      99%: 0.762079
Max Value: 0.897098
Expected Mean (0.075663) to be under expected maximum (0.065): FAIL
Expected 50% (0.371955) to be under expected maximum (0.29): FAIL
test msaa_line ... FAILED

failures:

failures:
    msaa_line

test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out; finished in 4.19s


--- STDERR:                                  wgpu-msaa-line-example::bin/msaa-line msaa_line ---
thread 'msaa_line' panicked at 'Image data mismatch!', /Users/kpreid/Projects/rust/contributing/wgpu/tests/src/image.rs:243:13
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
thread 'msaa_line' panicked at 'UNEXPECTED TEST FAILURE DUE TO PANIC', /Users/kpreid/Projects/rust/contributing/wgpu/tests/src/lib.rs:344:9

None of this tells me where to find the actual-output image to diagnose whether the image is wrong or the threshold is wrong.