Based on observations with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESAmember states and the USA (NASA).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

Appendices A and B are only available in electronic form at http://www.edpsciences.org

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... there

Owing to the finite CCD transfer time, a minor fraction of in FOV events is wrongly assigned to the out FOV region as Out Of Time events. Such events correspond to $\sim$ 0.35% of the out FOV surface brightness and therefore they may be neglected in our work.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... matrices

The ground calibration campaign demonstrated that the Quantum Efficiencies of the different CCDs are uniform above $\sim$ 0.3 keV (the differences were measured to be <2% at 2 keV, Vercellone 2000). In our analysis we account for possible systematics affecting the absolute flux measurement as discussed in Sect. 5.9.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.