It seems the point of the remark is to emphasize the symmetric property of the congruence relation i.e. $\,a\equiv b\iff b\equiv a\pmod m.\,$ This is not true for the operator mod, e.g we have $\,15\equiv 1\pmod 7\,$ and $\, (15\bmod 7) = 1\,$ and $\,1\equiv 15\pmod 7\,$ but $\ (1\bmod 7) \neq 15.\,$
The operational mod is antisymmetric since it converts its argument to normal form. The relational mod has no such preference, which often makes it more flexible in many contexts, e.g. $\,{\rm mod}\ 10\!:\,\ 9\equiv -1\,\Rightarrow\, 9^{2n}\equiv (-1)^{2n}\equiv ((-1)^2)^n\equiv 1^n\equiv 1,\,$ where we have used the flexibility of the congruence relation to choose a rep $\equiv 9\,$ which makes the power computation trivial.
Generally a congruence (to a fixed modulus) can be thought of as a generalized equality relation (i.e. an equivalence relation) that is, furthermore, consistent with the arithmetic operations (here obeys the Congruence Sum and Product Rules). This will be made more precise when you study quotient sets and algebras in abstract algebra.