Enable logging anchor profiling by default

[zlobober]

Re recent discussion in SIG MapReduce.

It would be really good to have the following option enable by default: https://github.com/ytsaurus/ytsaurus/blob/f781206ea03e38a6d777252c0de2e7ff51740a93/yt/yt/core/logging/config.cpp#L208-L209

It is often useful in situations when troubles are already happening, and you can see it in logs of individual components, but the corresponding profiling metric is not added to the code, so you cannot perceive the state of a cluster at scale.

I suggest turning it on by default, considering that it should not cost much on clusters of small-medium sizes (up to thousands of nodes), and the value of having it enabled as everywhere as possible is much higher than the inconvenience of one-time opting it out for the high-cardinality components (or, alternatively, tuning thresholds for them).