Is a meromorphic function always a ratio of two holomorphic functions?

1) Yes, every meromorphic function in $f\in \mathcal M(D)$ can be written $f=\frac{g}{h}$ with $g,h\in \mathcal O(D)$.
Proof: The divisor associated to $f$ has a canonical minimal decomposition $div(f)=D^+-D^-$ with $D^+-D^-\geq 0$ with $D^+, D^-$ effective (=positive)
[Explicitly: $D^+(z) =max(0, ord_zf), \; D^-(z) =max(0, -ord_zf)$ ]
The fundamental result (Rudin, Real and Complex Analysis, Theorem 15.11 page 295) is that there exists a function $h\in \mathcal O(D)$ with $div(h)=D^-$. If we then put $g=fh\in \mathcal O(D)$, we have our required decomposition as quotient $f=\frac{g}{h}$.

2) As a consequence, note the pleasant algebraic result $\mathcal M(D)=Frac(\mathcal O(D))$ characterizing the fraction field of the domain $\mathcal O(D)$.

3) The decomposition $f=\frac{g}{h}$ is also valid on any non-compact Riemann surface $X$ and again we have $\mathcal M(X)=Frac(\mathcal O(X))$.
The exact same proof as above works since any divisor $E\in Div(X)$ is the divisor of some meromorphic function $ m\in \mathcal M(X)^*$, that is $E=div(m)$: that result is proved in Theorem 26.5 of Forster's Lectures on Riemann Surfaces, the ultimate reference on the subject and a book I can't praise highly enough.

Yes; $M(D)$ is the quotient field of $H(D)$. Conway, "Functions of one complex variable", Corollary 5.20.