Installation: Don't insist on providing GMP (#8893)

## Summary of Changes

Removed error when GMP is not found.

## Release Management

* Affected package(s): Installation
* Issues: fix #8852

Installation/cmake/modules/CGAL_SetupGMP.cmake

@@ -21,9 +21,18 @@ set(CGAL_SetupGMP_included TRUE)
 # That is required to find the FindGMP and FindMPFR modules.
 set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} ${CGAL_MODULES_DIR})
 
-find_package(GMP REQUIRED)
+find_package(GMP QUIET)
-find_package(MPFR REQUIRED)
+find_package(MPFR QUIET)
-find_package(GMPXX QUIET)
+
+if (GMP_FOUND)
+  find_package(GMPXX QUIET)
+endif()
+
+if(NOT GMP_FOUND OR NOT MPFR_FOUND)
+  message(STATUS "GMP not found.")
+  set(CGAL_DISABLE_GMP ON)
+  return()
+endif()
 
 if(NOT GMPXX_FOUND)
   option(CGAL_WITH_GMPXX "Use CGAL with GMPXX: use C++ classes of GNU MP instead of CGAL wrappers" OFF)

Number_types/doc/Number_types/NumberTypeSupport.txt

@@ -115,6 +115,19 @@ requirements.
 
 To use these classes, \gmp and \mpfr must be installed.
 
+
+\section Number_typesBoost Number Types Provided by Boost
+
+\anchor boostnt
+
+Boost provides arbitrary precision integer and rational number types.
+The number types `boost::multiprecision::gmp_int` and `boost::multiprecision::gmp_rational`
+have \gmp as backend, while the number types `boost::multiprecision::cpp_int`
+and `boost::multiprecision::cpp_rational` have a native backend.
+The file <TT>CGAL/boost_mp.h</TT> provides the functions to make
+these classes models of number type concepts.
+
+
 \section Number_typesLEDA Number Types Provided by LEDA
 
 \anchor ledant

Number_types/doc/Number_types/PackageDescription.txt

@@ -9,6 +9,9 @@
 /// \defgroup nt_core CORE
 /// \ingroup PkgNumberTypesRef
 
+/// \defgroup nt_boost Boost
+/// \ingroup PkgNumberTypesRef
+
 /// \defgroup nt_leda LEDA
 /// \ingroup PkgNumberTypesRef
 
@@ -71,6 +74,13 @@
 - `leda_bigfloat`
 - `leda_real`
 
+\cgalCRPSubsection{Boost}
+
+- `boost::multiprecision::cpp_int`
+- `boost::multiprecision::cpp_rational`
+- `boost::multiprecision::gmp_int`
+- `boost::multiprecision::gmp_rational`
+
 \cgalCRPSubsection{GMP}
 
 - `mpz_class`

Number_types/include/CGAL/Exact_integer.h

@@ -27,8 +27,7 @@ namespace CGAL {
 `Exact_integer` is an exact integer number type.
 
 It is a typedef of another number type. Its exact definition depends on
-the availability the third-party libraries \gmp, \core, and \leda. \cgal must
+the availability the third-party libraries \gmp and \leda.
-be configured with at least one of those libraries.
 
 \cgalModels{EuclideanRing,RealEmbeddable}
 

Number_types/include/CGAL/Exact_rational.h

@@ -27,8 +27,7 @@ namespace CGAL {
 `Exact_rational` is an exact rational number type, constructible from `double`.
 
 It is a typedef of another number type. Its exact definition depends on
-the availability the third-party libraries \gmp, \core, and \leda. \cgal must
+the availability the third-party libraries \gmp and \leda.
-be configured with at least one of those libraries.
 
 \cgalModels{Field,RealEmbeddable,Fraction,FromDoubleConstructible}