Make apply_ssim_boost a const fn

[FreezyLemon]

The ILog trait was introduced in v_frame before {integer}::ilog became stable. const fn doesn't do much here, but it's simple to do, so one might as well.

Note that the functions are similar but not equal in behavior:
ILog::ilog is effectively

if x == 0 {
  0
} else {
  x.ilog2() + 1
}

x.ilog2() will panic if x == 0, but that is already assumed to not be the case here (and it looks like the check for zero is optimized away in all callsites).

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (3c3a26f) 88.52% compared to head (c7cef88) 88.54%.
Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3352      +/-   ##
==========================================
+ Coverage   88.52%   88.54%   +0.01%     
==========================================
  Files          89       89              
  Lines       28323    28355      +32     
==========================================
+ Hits        25074    25107      +33     
+ Misses       3249     3248       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

Hold on, I'm just double checking something here

[shssoichiro]

Just making sure, is the removal of the - 1 intentional here? My guess is yes given the docs in v_frame: floor(log2(self)) + 1, or 0 if self == 0, but want to be certain this is bit-identical to the prior output.

[FreezyLemon]

Yes, it's intentional and I did make sure they're equivalent for x != 0: https://rust.godbolt.org/z/Gh5TxbEYK