I'm not sure we can remove it, because even though it is not used for the means of the Android SDK, some users may rely on this as an exported symbol for their usage with com.datadog.trace.api.Tracer (this would be strange though).

Maybe it is better to deprecate it instead?

I really want to avoid the confusion here, I would say removing is better as there should not be any usage of the CoreTracer directly. This is our internal tracer and it is the only one that implements the com.datadog.trace.api.Tracer. If someone uses the CoreTracer directly then it will be a mistake.

Yes, for sure, its usage is a misuse, but at the same time it is binary-incompatible change and we can do it only for the next major release for the public-facing API (we cannot say that there is no anyone using this symbol for some strange needs).

At the same time java.lang.Deprecated doesn't allow to supply the message, but kotlin.Deprecated allows to do it (maybe we can use it in the Java code?).

But anyway, it is your call.

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 70.10%. Comparing base (661033b) to head (e18c280).
Report is 2 commits behind head on develop.

@@             Coverage Diff             @@
##           develop    #2438      +/-   ##
===========================================
+ Coverage    70.06%   70.10%   +0.04%     
===========================================
  Files          769      769              
  Lines        28511    28509       -2     
  Branches      4778     4778              
===========================================
+ Hits         19976    19986      +10     
+ Misses        7190     7185       -5     
+ Partials      1345     1338       -7     

I would tend to agree with @0xnm here: removing it might break using for some of our customers.
Adding a Deprecated annotation (and adding javadoc to explain why) seems like a better first step.
Awe could additionally add some telemetry to track the usage, and if we are sure no one uses it, we can then delete it.