Suppose we have a feasible LP problem in its standard form. From Ax=b we can directly determine some of its variables and thus we can reduce the problem.
For example, from two constraints: x+y+z=2 and x+y=1, we can determine z=1. Then we can replace 'z' with '1' and move it to the RHS, and delete variable 'z' as well as one of the constraints above.
Are there any tools which can do this kind of reduction?
The tools you are referring to are commonly called presolvers.
Every optimization software makes use of those (to improve performance, but also numerical-stability). This includes commercial solvers, as well as non-commercial ones.
Most implementations, if not all are embedded into the solver framework. The sole exception known to me would be:
Embedded candidates to look at:
Even if embedded, most most solvers allow to do the presolving step in an isolated step without automatically triggering follow-up solving (if the user wants it).
There are high-quality resources on this topic.
Andersen, Erling D., and Knud D. Andersen. "Presolving in linear programming." Mathematical Programming 71 (1995): 221-245.In this context, presolving is usually much more important and much more powerful (imho).
Achterberg, Tobias, et al. "Presolve reductions in mixed integer programming." INFORMS Journal on Computing 32.2 (2020): 473-506.Gamrath, Gerald, et al. "Progress in presolving for mixed integer programming." Mathematical Programming Computation 7 (2015): 367-398.