CoolProp: CoolProp can't load REFPROP v10 mixtures on Debian 9

Description

Corresponding REFPROP-cmake repo issue thread

I am installing REFPROP v10 and Python CoolProp v6.3.0 inside a docker container. To build REFPROP, I’m using Ian Bell’s cmake repo. The build executes successfully and I am able to use REFPROP through CoolProp in Python, but only for pure fluids. I am able to use mixtures if I run commands from the /opt/refprop folder where I have copied the REFPROP files, but nowhere else. Setting the ALTERNATIVE_REFPROP_PATH using CP.set_config_string(CP.ALTERNATIVE_REFPROP_LIBRARY_PATH, '/opt/refprop/') does not solve my issue.

Contents of my local Dockerfile:

FROM python:3.7 as build

# Install dependencies needed to build C extensions
RUN apt-get -y -m update && \
    apt-get install -y cmake g++ gfortran python-pip python3-dev

RUN pip install numpy six

WORKDIR /usr/local/refprop

# Copy in the refprop repo for building
COPY . .

# Make the default refprop location
RUN mkdir /opt/refprop

# Set path
ENV LD_LIBRARY_PATH="/opt/refprop:/usr/local/lib:$LD_LIBRARY_PATH"
ENV PATH="/opt/refprop:$PATH"
ENV RPPREFIX="/opt/refprop:/usr/local/lib:$RPPREFIX"

WORKDIR /usr/local/refprop/build

RUN cmake .. -DREFPROP_FORTRAN_PATH=/usr/local/refprop/files/fortran -DCMAKE_BUILD_TYPE=Release && \
    cmake --build .

# copy files to the default refprop locations
COPY files/fluids /opt/refprop/fluids/
COPY files/mixtures /opt/refprop/mixtures/
RUN cp -r /usr/local/refprop/build/librefprop.so /opt/refprop/librefprop.so
RUN cp -r /usr/local/refprop/build/librefprop.so /usr/local/lib/librefprop.so

WORKDIR /usr/local/refprop/

# Used for testing, doesn't hurt anything except image size
RUN pip install coolprop

Steps to Reproduce

1. Recursively clone the NIST REFPROP cmake repo: https://github.com/usnistgov/REFPROP-cmake
2. Copy above Dockerfile to repo root
3. Make a files folder in repo root
4. Copy FORTRAN, MIXTURES, FLUIDS REFPROP folders into files/ as lowercase directories (FORTRAN --> fortran)
5. run docker build -t refprop . from the repo root
6. Once build is completed, run docker run -it refprop bash to shell into the docker container
7. From /usr/local/refprop, open python

root@b050b185315f:/usr/local/refprop£ python
Python 3.7.2 (default, Dec 29 2018, 07:16:08) 
[GCC 6.3.0 20170516] on linux
Type "help", "copyright", "credits" or "license" for more information.

>>> import CoolProp.CoolProp as CP

>>> CP.PropsSI("H", "P", 1e5, "T", 300, "REFPROP::R407A.MIX")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "CoolProp/CoolProp.pyx", line 347, in CoolProp.CoolProp.PropsSI
  File "CoolProp/CoolProp.pyx", line 427, in CoolProp.CoolProp.PropsSI
  File "CoolProp/CoolProp.pyx", line 314, in CoolProp.CoolProp.__Props_err2
ValueError: Initialize failed for backend: "REFPROP", fluid: "R407A.MIX" fractions "[ 1.0000000000 ]"; error: Unable to load mixture: R407A.MIX : PropsSI("H","P",100000,"T",300,"REFPROP::R407A.MIX")

>>> CP.PropsSI("H", "P", 1e5, "T", 300, "REFPROP::water")
112653.67968889134

>>> CP.set_config_string(CP.ALTERNATIVE_REFPROP_LIBRARY_PATH, '/opt/refprop/')

>>> CP.PropsSI("H", "P", 1e5, "T", 300, "REFPROP::R407A.MIX")
Good news: It is possible to use REFPROP on your system! However, the library 
could not be loaded. Please make sure that REFPROP is available on your system.

Neither found in current location nor found in system PATH.
If you already obtained a copy of REFPROP from http://www.nist.gov/srd/, 
add location of REFPROP to the PATH environment variable or your library path.

In case you do not use Windows, have a look at https://github.com/jowr/librefprop.so 
to find instructions on how to compile your own version of the REFPROP library.

ALTERNATIVE_REFPROP_PATH: 
ERROR: Could not load REFPROP (/opt/refprop/) due to: /opt/refprop/: cannot read file data: Is a directory. Make sure the library is in your system search path. In case you run 64bit Windows and you have a REFPROP license, try installing the 64bit DLL from NIST. 
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "CoolProp/CoolProp.pyx", line 347, in CoolProp.CoolProp.PropsSI
  File "CoolProp/CoolProp.pyx", line 427, in CoolProp.CoolProp.PropsSI
  File "CoolProp/CoolProp.pyx", line 314, in CoolProp.CoolProp.__Props_err2
ValueError: Initialize failed for backend: "REFPROP", fluid: "R407A.MIX" fractions "[ 1.0000000000 ]"; error: You cannot use the REFPROPMixtureBackend. : PropsSI("H","P",100000,"T",300,"REFPROP::R407A.MIX")

>>> CP.PropsSI("H", "P", 1e5, "T", 300, "REFPROP::water")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "CoolProp/CoolProp.pyx", line 347, in CoolProp.CoolProp.PropsSI
  File "CoolProp/CoolProp.pyx", line 427, in CoolProp.CoolProp.PropsSI
  File "CoolProp/CoolProp.pyx", line 314, in CoolProp.CoolProp.__Props_err2
ValueError: Initialize failed for backend: "REFPROP", fluid: "water" fractions "[ 1.0000000000 ]"; error: You cannot use the REFPROPMixtureBackend. : PropsSI("H","P",100000,"T",300,"REFPROP::water")

1. Run Python from the refprop folder: cd /opt/refprop && python to see CoolProp work inside the REFPROP folder for both fluids and mixtures.

root@b050b185315f:/usr/local/refprop£ cd /opt/refprop/ && python
Python 3.7.2 (default, Dec 29 2018, 07:16:08) 
[GCC 6.3.0 20170516] on linux
Type "help", "copyright", "credits" or "license" for more information.

>>> import CoolProp.CoolProp as CP

>>> CP.PropsSI("H", "P", 1e5, "T", 300, "REFPROP::water")
112653.67968889134

>>> CP.PropsSI("H", "P", 1e5, "T", 300, "REFPROP::R407A.MIX")
428158.16100957023

Expected behavior:

CoolProp should have access to both pure and mixture REFPROP backends, not just pure fluids.

Actual behavior: CoolProp only has access to mixture REFPROP backends when commands are run from the /opt/refprop folder where the mixtures REFPROP folder is.

Versions

CoolProp Version: Python v6.3.0 Operating System and Version: Debian GNU/Linux 9 (stretch) (docker container: https://github.com/docker-library/python/blob/34c9df35e9a69e9f0edde88e861b543edb8bc07a/3.7/stretch/Dockerfile) Access Method: Python3.7

Additional Information

If possible, please post examples and/or screenshots of the issue.