Hashtable: Use upstream OpenJDK8u121-b13 versions of Map default methods

In Android N, the implementation of these methods was
built on the Map default implementation which in turn
was built on top of get(), put(), containKey() etc.
This can lead to behavior differences in subclasses of
Hashtable that have overridden only those methods and
rely on the details of how these methods were implemented
on top of those in Android. Such reliance on implementation
details would be inappropriate / not reasonable for an app
to do and is therefore not supported.

The new implementation (identical (*) to upstream) should
generally be faster, since it can make use of e.g. the
fact that Hashtable can not contain null keys or values.
However, Hashtable performance is not currently covered
by tests so this has not been empirically measured.

(*) Note that the upstream name for Hashtable.HashtableEntry
is Hashtable.Entry, so the added code that referred to
that class has been adjusted accordingly.

Test: CtsLibcoreTestCases -t org.apache.harmony.tests.java.util.HashtableTest
Test: CtsLibcoreTestCases -t libcore.java.util.HashtableTest
Test: CtsLibcoreTestCases
Bug: 35916491

Change-Id: Icddcfa442de6f2ca9f186fa7f489f9b9c0638d9a