[doc] Fix some typos.

doc/acas_xu.rst

@@ -346,7 +346,7 @@ We can then verify the resulting verification problem as before:
 
 It is interesting to remark that, since Marabou does not support disjunctive
 formulas, CAISAR first splits a disjunctive goal formula into conjunctive
-sub-goals, then calls Marabou on each sub-goals, and finally post-processes the
+sub-goals, then calls Marabou on each sub-goal, and finally post-processes the
 sub-results to provide the final result corresponding to the original goal
 formula.
 

doc/foreword.rst

@@ -29,7 +29,7 @@ The supported provers are the following:
   vector machine verifier based on abstract interpretation,
 * `nnenum <https://github.com/stanleybak/nnenum>`_ a neural network verifier
   that combines abstract interpretation and linear programming techniques,
-* *Classic* SAT/SMT solvers that support the SMT-LIBv2 input language.
+* Standard SAT/SMT solvers that support the SMT-LIBv2 input language.
 
 CAISAR aims to be component-agnostic: it can model and assess the
 trustworthiness of artificial intelligence system that potentially mixes both