CRT function replacement failures should be able to fail build

[sigurasg]

As-is, str* function replacement failure just logs, and then we get false-positive overruns from the field. For Chrome, it would be much preferable to fail the build, as that way we can at least fix it before annoying kN devs. Perhaps this could be then disabled under a flag (--dont-barf-on-crt-function-replacement-failure)?

...
[0615/231658:WARNING:asan_interceptor_filter.cc(65)] Not intercepting wcschr with hash 574d84de4f9718ae0d1e149aea4bef43
...
[0615/231658:WARNING:asan_interceptor_filter.cc(65)] Not intercepting strnlen with hash 96bece78e0fcd82e400ad92889100e0e
[0615/231658:WARNING:asan_interceptor_filter.cc(65)] Not intercepting wcsnlen with hash d39e85f57e04069cff34624893a84e4a
....

[sebmarchand]

I'm working on this. I'm not sure if the flag is needed as we already have a "no-interceptors" one?