To show there is a problem in your proof, I will try to make a similar proof, using another language. I hope you will find that the conclusion of my proof is incorrect and that you will see the similarities with yours. I will then try to explain in more details what the problem is.
Consider the language $L = \{a^mb^n\mid m,n\geqslant 0\}$. I assume that $L$ is regular. Since it is regular, it verifies the pumping lemma. Let then $p$ be the constant in the pumping lemma, and $s = ab^p$. Since $s\in L$ and $s$ has length greater than $p$, then $s$ can be split into $s = xyz$ where for $i\geqslant 0$, $xy^iz\in L$.
I choose the following split to prove a contradiction: divide $s$ into $x = \varepsilon$, $y = ab$ and $z = b^{p-1}$. In this case, $xyyz = abab^p$ is not a word of $L$ therefore $L$ is not regular.
Now obviously there is a problem with the conclusion because $L = a^*b^*$ is clearly a regular language.
To see the problem, let's do a bit of logic. The pumping lemma states:
Being regular $\Rightarrow$ Verifying pumping lemma conclusion
Therefore, we conclude (with contraposition):
Not verifying pumping lemma conclusion $\Rightarrow$ Not being regular
This is the property used to prove a language is not regular. Let's look a little bit closer to the pumping lemma conclusion:
$\exists p$ such that, $\forall s$ of length $\geqslant p$, $\exists xyz$ verifying the three properties
The negation of that conclusion is therefore:
$\forall p$, $\exists s$ of length $\geqslant p$, $\forall xyz$, the three properties are not verified.
That's why I insist that you need to consider ALL $xyz$ decompositions.
Now to go back to Sipser's proof, contrary to what you seem to say in an above comment, I do not claim that the proof is wrong, I claim that your interpretation of it is wrong. Indeed, Sipser says that for a fixed $p$, we consider $s = 0^p1^p$. He then consider a decomposition where $y$ consists only of $0$'s. But he does that because, implicitly, he considers that the decomposition verifies the properties 2. and 3., meaning that $|y| > 0$ and $|xy| < p$. Since $|xy| < p$, that means $xy$ is included in the $0^p$ part of $s$.
By doing that, he is still considering all decomposition of $s$, but is implicitly discarding those that do not verify 2. or 3. (because those ones are already missing some properties needed for the conclusion of the pumping lemma).