Use vfsgen instead of go-bindata

[lunny]

The embed package github.com/shurcooL/vfsgen has been vendored and will be invoke on go generate ./modules/.... So gitea build process will not be broken even the embed package API changed or not maintained. This should fix #4250.

@lafriks @tboerger maybe you can give some ideas.

@lafriks Just noticed your comment at #4250 (comment) .

[tboerger]

I personally use gorunpkg to execute scripts from modules instead of adding this kind of run wrapper :)

[lunny]

Script will be only invoked on mac/linux, but a go wrapper will be called much platforms.

[codecov-io]

Codecov Report

Merging #7080 into master will decrease coverage by <.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #7080      +/-   ##
==========================================
- Coverage   41.54%   41.53%   -0.01%     
==========================================
  Files         446      446              
  Lines       60825    60825              
==========================================
- Hits        25267    25261       -6     
- Misses      32273    32280       +7     
+ Partials     3285     3284       -1

Impacted Files	Coverage Δ
modules/options/options.go	96.15% <ø> (ø)	⬆️
modules/public/public.go	74.41% <ø> (ø)	⬆️
models/unit.go	62.16% <0%> (-5.41%)	⬇️
routers/repo/view.go	42.02% <0%> (-1.02%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8eba27c...89e1e29. Read the comment docs.

[lunny]

It's ready to review.

[lunny]

I noticed that this PR will increase gitea start memory from 64MB to 81MB on macOS. Should I enable some compress?

[zeripath]

@lunny if you think it will help - sometimes compression can actually even speed things up!

[lunny]

@zeripath yes, for public resouces(except images), we could directly provide compressed bytes to browser. Currently vsfgen will always save compressed bytes on memory with a map. So for a HTTP request, it will be decompress and compress. This will waste some CPUs. I will send another PR to resolve that.