v7.3.1

Adds a secondary computation of primal and dual feasibility errors using double precision (in addition to the feasibility errors computed using arbitrary precision).

Many practical applications operate within double precision tolerances, and there is some value in reporting feasibility errors using double precision.

v7.3.0

Besides minor changes, this release is primarily meant to bump the version number to v7.3.0 from v0.1.0 (the pre-release in April 2023).

SDPA Multiprecision forked off from SDPA-GMP 7.1.3. The following features (from the regular SDPA 7.3.16 package) were incorporated in the April 2023 prerelease:

1. Usage of sdpa_call.{h,cpp} and sdpa_solve.cpp as the backbone of the program. This allows SDPA Multiprecision to be used as a library through sdpa_call.h and libsdpa_gmp.a static library.
2. sdpa_exe.cpp provides an executable frontend and replaces the earlier sdpa_main.cpp.
3. Allowing custom print precision of the primal and dual variables by setting xPrint, XPrint and YPrint in param.sdpa file. Please read the documentation of GNU Multiple Precision Library to set these properly.
4. Usage of BlockStruct class (defined in sdpa_block.{h,cpp}) to reduce redundancy of code.

The addition of the ability to be used as libsdpa_gmp, brings it at par with the main SDPA package by functionality (currently SDPA 7.3.17). Since semantic versioning follows major.minor.patch syntax (and major.minor components are concerned with API and functionality), the version number is being bumped to 7.3.0 to match major.minor of the current SDPA 7.3.17.