[BUG] Latest code require boost 1.74 but should only require boost 1.72

[esoha-nvidia]

Describe the bug
A nightly version of cudf inadvertantly depended on boost 1.74. This was reverted and latest cudf is now back to depending upon boost 1.72. In the meantime, a blazingsql nightly was created that used that mistaken cudf and the new dependency clashes with the latest cudf.

Steps/Code to reproduce bug
The list of conda releases here is used to generate the two URLs below.

1. Examine the dependencies of the latest nightly and notice that it requires boost 1.74.

   wget 'https://anaconda.org/blazingsql-nightly/blazingsql/0.16.0a/download/linux-64/blazingsql-0.16.0a-cuda10.2_py37_52.tar.bz2'
   tar xvjf blazingsql-0.16.0a-cuda10.2_py37_52.tar.bz2
   ldd lib/libblazingsql-engine.so  | grep boost
   

2. Examine the dependencies of the previous nightly and notice that it requires boost 1.72.

   wget 'https://anaconda.org/blazingsql-nightly/blazingsql/0.16.0a/download/linux-64/blazingsql-0.16.0a-cuda10.2_py37_25.tar.bz2'
   tar xvjf blazingsql-0.16.0a-cuda10.2_py37_25.tar.bz2
   ldd lib/libblazingsql-engine.so  | grep boost
   

Expected behavior
blazingsql should continue to depend on boost 1.72

Environment overview (please complete the following information)

• Method of cuDF install: conda

[wmalpica]

Thanks @esoha-nvidia a new conda package is available that no longer requires boost 1.74