Okay:
f(x)\rightarrow f(p_{f})=\frac{p_{f}^{2}}{2m}-E_{0}^{i}-\hbar\omega (6)
Solving the equation
f(p_{f})=0 (7)
,yields the 2 solutions (which fortunately have the degree of multiplicity exactly 1)
p_{f}^{1,2}=\pm \sqrt{2m(E_{0}^{i}+\hbar\omega)} (8)
Computing the derivative of the function on the solutions (8) of the equation (7),we get,after considering the modulus/absolute value:
\frac{df(p_{f})}{dp_{f}}=\frac{\sqrt{2m(E_{0}^{i}+\hbar\omega)}}{m} (9)
Combining (8),(9) and the general formula (5) (v.prior post),we get:
\delta (\frac{p_{f}^{2}}{2m}-E_{0}^{i}-m_{i})=\frac{m}{\sqrt{2m(E_{0}^{i}+\hbar\omega)}} \{\delta[p_{f}-\sqrt{2m(E_{0}^{i}+\hbar\omega)}]+\delta[p_{f}+\sqrt{2m(E_{0}^{i}+\hbar\omega)}]\} (10)
which is totally different than what the OP had posted...
IIRC,when learning QFT,i always said to myself:theorem of residues and the theory of distributions go hand in hand...
Daniel.