GhcDynamic build system variable is no longer supported; but we have references to it

ghc/ghc.mk currently rejects GhcDynamic:

ifneq "$(GhcDynamic)" ""
$(error GhcDynamic is no longer supported, use DYNAMIC_GHC_PROGRAMS instead)
endif

However a quick search reveals that we check this variable in the build system and in some of the tests:

$ ag -w "GhcDynamic"
testsuite/driver/testlib.py
146:#     quasiquotation/qq007/Makefile:ifeq "$(GhcDynamic)" "YES"

testsuite/tests/quasiquotation/qq007/Makefile
7:ifeq "$(GhcDynamic)" "YES"

testsuite/tests/quasiquotation/qq009/Makefile
7:ifeq "$(GhcDynamic)" "YES"

testsuite/tests/quasiquotation/qq008/Makefile
7:ifeq "$(GhcDynamic)" "YES"

testsuite/mk/ghc-config.hs
27:  getGhcFieldOrDefault fields "GhcDynamic" "GHC Dynamic" "NO"

testsuite/mk/test.mk
167:ifeq "$(GhcDynamic)" "YES"

testsuite/mk/boilerplate.mk
268:# code that uses TemplateHaskell with -prof, while GhcDynamic=YES.
269:ifeq "$(GhcDynamic)" "YES"

hadrian/src/Oracles/TestSettings.hs
61:        TestGhcDynamic            -> "GhcDynamic"

ghc/ghc.mk
56:ifneq "$(GhcDynamic)" ""
57:$(error GhcDynamic is no longer supported, use DYNAMIC_GHC_PROGRAMS instead)

I'm not sure if these uses of GhcDynamic should be replaced with DYNAMIC_GHC_PROGRAMS or should just be gone (assume that it'll never be set).