Add boostmp to test matrix, also bump llvm version tested.

[bjodah]

When building symengine with boostmp, I get a test failure in symengine.py's test suite. This is due to integer division in boost's multiprecision library throws, and we never catch the c++ excpetion, leading to the exit of the CPython process. I suspect we need to pass a policy somewhere, but first I want to reproduce the error in our CI run.

TODO:

• fix our use of boost's mulitprecision integer so that it does not throw for division-by-zero.

Stack trace:

gdb backtrace

tests/test_ntheory.py::test_divides                                                                      
Thread 1 "python3" received signal SIGABRT, Aborted.                                                     
__pthread_kill_implementation (no_tid=0, signo=6, threadid=140737350246400) at ./nptl/pthread_kill.c:44  
44      ./nptl/pthread_kill.c: No such file or directory.                                                
(gdb) bt                                                                                                 
#0  __pthread_kill_implementation (no_tid=0, signo=6, threadid=140737350246400) at ./nptl/pthread_kill.c:
44                                                                                                       
#1  __pthread_kill_internal (signo=6, threadid=140737350246400) at ./nptl/pthread_kill.c:78              
#2  __GI___pthread_kill (threadid=140737350246400, signo=signo@entry=6) at ./nptl/pthread_kill.c:89      
#3  0x00007ffff7c8d476 in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26                    
#4  0x00007ffff7c737f3 in __GI_abort () at ./stdlib/abort.c:79                                           
#5  0x00007ffff41eebbe in ?? () from /lib/x86_64-linux-gnu/libstdc++.so.6                                
#6  0x00007ffff41fa24c in ?? () from /lib/x86_64-linux-gnu/libstdc++.so.6                                
#7  0x00007ffff41fa2b7 in std::terminate() () from /lib/x86_64-linux-gnu/libstdc++.so.6                  
#8  0x00007ffff41fa518 in __cxa_throw () from /lib/x86_64-linux-gnu/libstdc++.so.6                       
#9  0x00007ffff4c905c4 in boost::throw_exception (e=..., loc=...) at /usr/include/bo
ost/throw_exception.hpp:171                                                                              
#10 0x00007ffff4caa75a in boost::multiprecision::backends::divide_unsigned_helper >, boost::multiprecision::backends::cpp_int_backend
<0u, 0u, (boost::multiprecision::cpp_integer_type)1, (boost::multiprecision::cpp_int_check_type)0, std::a
llocator > > (result=0x0, x=..., y=0, r=...) at /usr/include/boost/multiprecision/cpp
_int/divide.hpp:339                                                                                      
#11 0x00007ffff4c9fd3e in boost::multiprecision::backends::divide_unsigned_helper >, boost::multiprecision::backends::cpp_int_backend
<0u, 0u, (boost::multiprecision::cpp_integer_type)1, (boost::multiprecision::cpp_int_check_type)0, std::a
llocator >, boost::multiprecision::backends::cpp_int_backend<0u, 0u, (boost::multipre
cision::cpp_integer_type)1, (boost::multiprecision::cpp_int_check_type)0, std::allocator > > (result=0x0, x=..., y=..., r=...) at /usr/include/boost/multiprecision/cpp_int/divide.hpp:73     
#12 0x00007ffff4d7f2c0 in boost::multiprecision::backends::eval_modulus<0u, 0u, (boost::multiprecision::c
pp_integer_type)1, (boost::multiprecision::cpp_int_check_type)0, std::allocator, 0u, 
0u, (boost::multiprecision::cpp_integer_type)1, (boost::multiprecision::cpp_int_check_type)0, std::alloca
tor, 0u, 0u, (boost::multiprecision::cpp_integer_type)1, (boost::multiprecision::cpp_
int_check_type)0, std::allocator > (result=..., a=..., b=...) at /usr/include/boost/m
ultiprecision/cpp_int/divide.hpp:532                                                                     
#13 0x00007ffff4effcf4 in boost::multiprecision::operator% > > (a=..., b=...) at /usr/include/boost/multiprecision/detail/no_et_ops.h
pp:176                                                                                                   
#14 0x00007ffff4f97908 in SymEngine::mp_divisible_p (a=..., b=...) at //symengine-d21cfe1ed57168f2b4e0d5b
e854632814387a3f1/symengine/mp_class.h:993                                                               
#15 0x00007ffff4f88c44 in SymEngine::divides (a=..., b=...) at /symengine-d21cfe1ed57168f2b4e0d5be8546328
14387a3f1/symengine/ntheory.cpp:163                                                                      
#16 0x00007ffff570e5a7 in __pyx_pf_9symengine_3lib_17symengine_wrapper_138divides (__pyx_self=0x0, __pyx_
v_a=0x7ffff78600d0, __pyx_v_b=0x7ffff78600d0) at /opt/symengine.py-74867351ed7ffb2f2e97f5807976b54953c0b7
1d/build/lib.linux-x86_64-cpython-310/symengine/lib/symengine_wrapper.cpp:110638

the exception being thrown

#10 0x00007ffff4caa75a in boost::multiprecision::backends::divide_unsigned_helper >, boost::multiprecision::backends::cpp_int_backend<0u, 0u, (boost::multiprecision::cpp_integer_type)1, (boost::multiprecision::cpp_int_check_type)0, std::allocator > > (result=0x0, x=..., y=0, r=...) at /usr/include/boost/multiprecision/cpp_int/divide.hpp:339
339           BOOST_THROW_EXCEPTION(std::overflow_error("Integer Division by zero."))

[bjodah]

I hope a self merge was OK here (nothing too controversial), this brings master (more) up to date with (c++) symengine.