Use std::numeric_limits<double>::max() instead of inf()

[artv3]

I noticed sometimes we run into floating point issues using inf(). Could we consider using max() instead?

[vladotomov]

yes