This should now properly handle small numbers which would previously
output something like `[0.00000000000000019, 0.00000000000000019]`,
which does not allow user to read the numbers properly.

Closes #1760

Only works for exceptions that publicly derive from `std::exception`
and the matching is done exactly, including case and whitespace.

Closes #1649
Closes #1728

# Please enter the commit message for your changes. Lines starting
# with '#' will be ignored, and an empty message aborts the commit.
# On branch master
# Your branch is up-to-date with 'origin/master'.
#
# Changes to be committed:
#	modified:   ../docs/matchers.md
#	modified:   ../include/internal/catch_capture_matchers.h
#	modified:   ../projects/CMakeLists.txt
#	modified:   ../projects/SelfTest/Baselines/compact.sw.approved.txt
#	modified:   ../projects/SelfTest/Baselines/console.std.approved.txt
#	modified:   ../projects/SelfTest/Baselines/console.sw.approved.txt
#	modified:   ../projects/SelfTest/Baselines/junit.sw.approved.txt
#	modified:   ../projects/SelfTest/Baselines/xml.sw.approved.txt
#	modified:   ../projects/SelfTest/UsageTests/Matchers.tests.cpp
#
# Untracked files:
#	./
#	../clang-full/
#	../clang-test/
#	../clang10-build/
#	../coverage-build/
#	../gcc-build/
#	../gcc-full/
#	../include/internal/catch_matchers_exception.cpp
#	../include/internal/catch_matchers_exception.hpp
#	../misc-build/
#	../msvc-sln/
#	../notes.txt
#	../test-install/
#

It checks Knuth's _close enough with tolerance_ relationship, that
is `|lhs - rhs| <= epsilon * max(|lhs|, |rhs|)`, rather then the
_very close with tolerance_ relationship that can be written down as
`|lhs - rhs| <= epsilon * min(|lhs|, |rhs|)`.

This is because it is the more common model around the internet, and
as such is likely to be less surprising to the users. In the future
we might want to provide the other model as well.

Closes #1746

Previously it was limited to roughly 2 billion ULPs, rather than
the roughly 2^64 possible ones.

Closes #1581

No matcher actually uses it, and there is no good reason for it,
as the best it can do for user is removing a single indirection
when using the pointer inside the matcher. Given the overhead of
other code that will be running during such time, it is completely
meaningless.

This also fixes compilation for PredicateMatcher<const char*>.

Previously a mismatched prefix would be skipped before the actual
comparison would be performed. Obviously, it is supposed to be
_matching_ prefix that is skipped.