Compile warnings on clang >= 6

[jeromekelleher]

Because of an upstream bug in clang >= 6, we get warnings with floating point operations like isnan and isfinite,

FAILED: tskit@sha/tskit_trees.c.o                               
clang -Itskit@sha -I. -I.. -Isubprojects/kastore -I../subprojects/kastore -Xclang -fcolor-diagnostics -pipe -D_FILE_OFFSET_BITS=64 -Wall -Winvalid-pch -std=c99 -g -fPIC -Wall -Wextra -Werror
 -Wpedantic -W -Wmissing-prototypes -Wstrict-prototypes -Wconversion -Wshadow -Wpointer-arith -Wcast-align -Wcast-qual -Wwrite-strings -Wnested-externs -fshort-enums -fno-common -MD -MQ 'tsk
it@sha/tskit_trees.c.o' -MF 'tskit@sha/tskit_trees.c.o.d' -o 'tskit@sha/tskit_trees.c.o' -c ../tskit/trees.c
../tskit/trees.c:284:28: error: implicit conversion loses floating-point precision: 'const double' to 'float' [-Werror,-Wconversion]
                if (!isnan(mutation_time[mutation])              
                     ~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~                                                                             
/usr/include/math.h:644:46: note: expanded from macro 'isnan'
#  define isnan(x) __MATH_TG ((x), __isnan, (x))                                                                                
                   ~~~~~~~~~~~~~~~~~~~~~~~~~~^~~                
/usr/include/math.h:559:16: note: expanded from macro '__MATH_TG'
   ? FUNC ## f ARGS                             \               
     ~~~~~~~~~ ^~~~                                              
1 error generated.                                  
[8/52] Compiling C object 'tskit@sha/tskit_tables.c.o'.          
FAILED: tskit@sha/tskit_tables.c.o                                                                                              
clang -Itskit@sha -I. -I.. -Isubprojects/kastore -I../subprojects/kastore -Xclang -fcolor-diagnostics -pipe -D_FILE_OFFSET_BITS=64 -Wall -Winvalid-pch -std=c99 -g -fPIC -Wall -Wextra -Werror
 -Wpedantic -W -Wmissing-prototypes -Wstrict-prototypes -Wconversion -Wshadow -Wpointer-arith -Wcast-align -Wcast-qual -Wwrite-strings -Wnested-externs -fshort-enums -fno-common -MD -MQ 'tsk
it@sha/tskit_tables.c.o' -MF 'tskit@sha/tskit_tables.c.o.d' -o 'tskit@sha/tskit_tables.c.o' -c ../tskit/tables.c
../tskit/tables.c:6592:23: error: implicit conversion loses floating-point precision: 'double' to 'float' [-Werror,-Wconversion]
        if (!isfinite(node_time)) {                             
             ~~~~~~~~~^~~~~~~~~~                                 
/usr/include/math.h:629:50: note: expanded from macro 'isfinite'
#  define isfinite(x) __MATH_TG ((x), __finite, (x))             
                      ~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~                                                                            
/usr/include/math.h:559:16: note: expanded from macro '__MATH_TG'
   ? FUNC ## f ARGS                             \                                                                               
     ~~~~~~~~~ ^~~~                                             
../tskit/tables.c:6634:24: error: implicit conversion loses floating-point precision: 'double' to 'float' [-Werror,-Wconversion]
        if (!(isfinite(left) && isfinite(right))) {             
              ~~~~~~~~~^~~~~                                     
/usr/include/math.h:629:50: note: expanded from macro 'isfinite'
#  define isfinite(x) __MATH_TG ((x), __finite, (x))             
                      ~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~
/usr/include/math.h:559:16: note: expanded from macro '__MATH_TG'
   ? FUNC ## f ARGS                             \                                                                               
     ~~~~~~~~~ ^~~~                               
../tskit/tables.c:6634:42: error: implicit conversion loses floating-point precision: 'double' to 'float' [-Werror,-Wconversion]
        if (!(isfinite(left) && isfinite(right))) {             
                                ~~~~~~~~~^~~~~~     

This is a problem with clang, see e.g. https://bugs.llvm.org/show_bug.cgi?id=35268