[mpfr] Fix when building shared library on Linux

[uilianries]

Specify library name and version: mpfr/4.2.0

When building with Conan 2.0, and */*:shared=True, the mpfr executes a validation with autotools to check if is possible to use gmp. However, as VirtualRunEnv was not configured, the LD_LIBRARY_PATH was not present, resulting in failure when trying to run its test:

configure:4522: gcc -o conftest  -m64 -O3  -I/home/conan/.conan2/p/gmp0bfdd57764b0f/p/include -DNDEBUG -I/home/conan/.conan2/p/gmp0bfdd57764b0f/p/include  -L/home/conan/.conan2/p/gmp0bfdd57764b0f/p/lib -m64 -L/home/conan/.conan2/p/gmp0bfdd57764b0f/p/lib conftest.c  -lgmpx
x -lgmp -lm >&5
configure:4526: $? = 0
configure:4533: ./conftest
./conftest: error while loading shared libraries: libgmpxx.so.4: cannot open shared object file: No such file or directory
configure:4537: $? = 127
configure:4544: error: in `/home/conan/.conan2/p/b/mpfrbaa826e03f170/b/build-release':
configure:4546: error: cannot run C compiled programs.

This PR fixes the situation by adding VirtualRunEnv.

Required by #18720 and #19124

---

• I've read the contributing guidelines.
• I've used a recent Conan client version close to the currently deployed.
• I've tried at least one configuration locally with the conan-center hook activated.

[conan-center-bot]

Conan v1 pipeline ✔️

All green in build 1 (fd6e77d9ba7e145003016a5b11dba5450f25d27a):

• mpfr/4.1.0@:
  All packages built successfully! (All logs)

• mpfr/4.0.2@:
  All packages built successfully! (All logs)

• mpfr/4.2.0@:
  All packages built successfully! (All logs)

---

Conan v2 pipeline ✔️

Note: Conan v2 builds may be required once they are on the v2 ready list

All green in build 1 (fd6e77d9ba7e145003016a5b11dba5450f25d27a):

• mpfr/4.1.0@:
  All packages built successfully! (All logs)

• mpfr/4.2.0@:
  All packages built successfully! (All logs)

• mpfr/4.0.2@:
  All packages built successfully! (All logs)

[jcar87]

We need to find a way to solve this that doesn't require exposing the run environment at build time!

[nextsilicon-itay-bookstein]

Writing it out here for posterity: apparently this trick causes libgmp from the conan cache to load into the compiler process itself! I only found this out as I was doing an address-sanitized build (with shared-libasan) and this caused weird errors to pop up when the configure script tried interrogating the compiler itself (missing symbols that normally would be retrieved from libasan DSO).