A hyper luminous starburst at z=4.72 magnified by a lensing galaxy pair at z=1.48

[Abridged] We discovered in the Herschel Reference Survey an extremely bright IR source with $S_{500}$~120mJy (Red Virgo 4 - RV4). Based on IRAM/EMIR and IRAM/NOEMA detections of the CO(5-4), CO(4-3), and [CI] lines, RV4 is located at z=4.724, yielding a total observed L$_{IR}$ of 1.1+/-0.6x0$^{14}$L$_{odot}$. At the position of the Herschel emission, three blobs are detected with the VLA at 10cm. The CO(5-4) line detection of each blob confirms that they are at the same redshift with the same line width, indicating that they are multiple images of the same source. In Spitzer and deep optical observations, two sources, High-z Lens 1 (HL1) West and HL1 East, are detected at the center of the three VLA/NOEMA blobs. These two sources are placed at z=1.48 with XSHOOTER spectra, suggesting that they could be merging and gravitationally lensing the emission of RV4. HL1 is the second most distant lens known to date in strong lensing systems. The Einstein radius of the lensing system is 2.2+/-0.2 (20kpc). The high redshift of HL1 and the large Einstein radius are highly unusual for a strong lensing system. We present the ISM properties of the background source RV4. Different estimates of the gas depletion time yield low values suggesting that RV4 is a SB galaxy. Among all high-z SMGs, this source exhibits one of the lowest L$_{[CI]}$ to L$_{IR}$ ratios, 3.2+/-0.9x10$^{-6}$, suggesting an extremely short gas tdepl of only 14+/-5Myr. It also shows a relatively high L$_{[CI]}$ to L$_{CO(4-3)}$ ratio (0.7+/-0.2) and low L$_{CO(5-4)}$ to L$_{IR}$ ratio (only ~50% of the value expected for normal galaxies) hinting a low density of gas. Finally, we discuss that the short tdepl of RV4 can be explained by either a very high SFE, which is difficult to reconcile with major mergers simulations of high-z galaxies, or a rapid decrease of SF, which would bias the estimate of tdepl toward low value.