Math spacing in TeX Gyre Pagella

The spacing in the following output looks off to me. In particular, the integral symbol has not grown to accommodate the height of the integrand, the spacing in the fraction seems large, and the enclosing brackets do not rise high enough in the matrix. The issue remains whether mathtools is loaded or not, but since I use it, I thought that a solution (if one exists) should work with the package.

I'm wondering if this is by design, or whether I've loaded the packages correctly/something is peculiar with my setup. If it is by design, is there a parameter that can be altered to change the spacing globally?

\documentclass{article}

\usepackage{mathtools}
\usepackage{luatextra}
\usepackage{unicode-math}
\setmathfont{TG Pagella Math}

\begin{document}

$\int_{x=0}^{\infty}\frac{1}{x}\,\mathrm dx$
$\begin{bmatrix*}1&1&1\\1&1&1\\1&1&1\end{bmatrix*}$

\end{document}
-
The integral symbol never stretches to cover the integrand; it's true that the symbol in TeX Gyre Pagella Math is rather small. – egreg Mar 26 '13 at 17:01
@egreg Ah, thanks for pointing that out...I had never noticed so just assumed that it did. – Scott H. Mar 26 '13 at 17:06
The brackets seem to be badly defined: the top is too low. – egreg Mar 26 '13 at 17:11
Have you tried with mathpazo and pdflatex? – pml Mar 26 '13 at 18:41
@pml I haven't but I'll give it a shot. I am still curious about the above however. – Scott H. Mar 26 '13 at 19:46

Sorry my decrepit home system doesn't auto-upgrade packages, and so I had to turn off the key packages which produce your problem. Nonetheless, the solution I offer should work with those packages enabled, though the parameters may need to be tweaked.

If you wished to change the size of your integral sign (not endorsed by egreg, I realize), the scalerel package can do it for you. I show your two equations and then I show two versions of scalerel, the first in which the integral is scaled to a specified size, and the second where it is scaled to match the size of an argument (in this case, your matrix [and yes, I realize the differential is not well formed, mathematically]). With both types of scales, one can issue as an optional argument a width limiter, which I exercised in both cases.

EDITED (JUN2015) to correct a backward incompatibility with earlier scalerel version.

\documentclass{article}

\usepackage{scalerel}
\usepackage{mathtools}
%\usepackage{luatextra}
%\usepackage{unicode-math}
%\setmathfont{TG Pagella Math}

\begin{document}

$\int_{x=0}^{\infty}\frac{1}{x}\,\mathrm dx$
$\begin{bmatrix*}1&1&1\\1&1&1\\1&1&1\end{bmatrix*}$
${\scaleto[2ex]{\int}{6ex}}_{\!\!x=0}^{\infty}\frac{1}{x}\,\mathrm dx$

\def\x{\begin{bmatrix*}1&1&1\\1&1&1\\1&1&1\end{bmatrix*}}
${\scalerel*[2.2ex]{\int}{\x}}_{\!\!x=0}^{\infty}\x$

\end{document}

-
Thanks for the answer Steven. This may be an option if there's no global adjustment mechanism. – Scott H. Mar 27 '13 at 5:12