For your first and third problem both Solve and Reduce are convinced there are no solutions.
For your second problem Solve cannot crack it, but Reduce is able to do it
TM={{Cos[phi]*Cos[si]-Cos[theta]*Sin[phi]*Sin[si], Cos[si]*Sin[phi]+Cos[theta]*Cos[phi]*Sin[si], Sin[theta]*Sin[si]},
{-Cos[theta]*Cos[si]*Sin[phi]-Cos[phi]*Sin[si], -Sin[phi]*Sin[si]+Cos[theta]*Cos[phi]*Cos[si], Cos[si]*Sin[theta]},
{Sin[theta]*Sin[phi], -Cos[phi]*Sin[theta], Cos[theta]}};
Simplify[Reduce[TM.{1, 0, 0}=={1, 0, 0} && 0<=phi<2Pi &&
-Pi/2<=theta<Pi/2 && 0<=si<2Pi,{phi, theta, si}],
0<=phi<2Pi && -Pi/2<=theta<Pi/2 && 0<=si<2Pi]
which quickly returns
(phi == 0 && si == 0) ||
(phi == Pi && si == Pi) ||
(theta == 0 &&
((si + 2*ArcTan[Tan[phi/2]] == 2*Pi && 0<phi<Pi]) ||
(si == -2*ArcTan[Tan[phi/2]] && phi > Pi)))
Checking each of those potential solutions, and many more, all return True
TM.{1, 0, 0}=={1, 0, 0}/.{phi->0, si->0}
TM.{1, 0, 0}=={1, 0, 0}/.{phi->Pi, si->Pi}
TM.{1, 0, 0}=={1, 0, 0}//.{theta->0,phi->Pi/2, si->2 Pi-2*ArcTan[Tan[phi/2]]}
TM.{1, 0, 0}=={1, 0, 0}//.{theta->0,phi->Pi/8, si->2 Pi-2*ArcTan[Tan[phi/2]]}
TM.{1, 0, 0}=={1, 0, 0}//.{theta->0,phi->3Pi/2, si->-2*ArcTan[Tan[phi/2]]}
TM.{1, 0, 0}=={1, 0, 0}//.{theta->0,phi->9Pi/8, si->-2*ArcTan[Tan[phi/2]]}
If your configuration implies there must be a unique solution for each of those problems then the only thing I can imagine is that some flaw has crept into the the construction of the problem.