Try out bigger inline threshold in clang release

[danlark1]

I hereby agree to the terms of the CLA available at: https://yandex.ru/legal/cla/?lang=en

Changelog category (leave one):

• Performance Improvement

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):
Try out clang inline threshold 1000

Detailed description / Documentation draft:
Let's try out inline threshold 1000 to see the binary size, compilation times and performance gains. Also it might show some places where inlining is crucial and we can manually write force inlines.

[danlark1]

Nothing interesting, let's try out experimental-new-pass-manager

[danlark1]

ThinLTO: 0.5%
ThinLTO + fexperimental-new-pass-manager: 1%, 2 min compile time reduction + 50mb smaller binary
ThinLTO + fexperimental-new-pass-manager + inline-threshold=1000: 4%, 7 min compile time increase + 450mb bigger binary
ThinLTO + fexperimental-new-pass-manager + inline-threshold=500: 2%, 3 min compile time increase + 170mb bigger binary

[alexey-milovidov]

4% will pay for 7 minutes and +450 MB binary size :)

[alexey-milovidov]

PS. I cannot understand the reason for MSan error :(

[danlark1]

PS. I cannot understand the reason for MSan error :(

100% unrelated flap