A standard barrel is 42 gallons. A relic of the 19th century days when oil from Titusville went east in 50 gallon whiskey barrels (also a notable product of western Pennsylvania) but part evaporated en route.
Anyway, refinery gain is only an issue because we measure oil in volume terms instead of mass (tonnes). Because of the law of the conservation of matter, a tonne of crude oil going into a refinery will produce a tonne of products (including processing losses along the way). But a barrel of crude oil going into a refinery might produce 1.1 barrels coming out. This is because lighter products (gasoline, kerosene, diesel, naphtha) are less dense than the crude they are made from and thus take up more volume for the same amount of weight. For example, there's typically 7.3 barrels of a medium crude to a tonne, but 7.8 barrels of diesel to a tonne, 8.5 of gasoline, and 11.65 barrels of LPG in a tonne.
As for end-use of oil, those data are the most difficult to compile. The graphic above is just the general yield of a barrel of oil and what it is generally used for, not the actual proportion in end-use. The IEA compiles these end-use statistics at the level of transport, agriculture, industry, feedstock use, etc. (i.e. sectoral end-use), as do national governments. A big difficulty in compiling world wide figures is that some countries (notably China) do not collect end-use data on the same basis as other countries. For example, China allocates gasoline end-use to the sector of the economy that consumes it, so that you end up with data such as "industrial use of gasoline", which is in effect, the use of gasoline for transport by entities that are considered industrial. In the US, all such use is just defined as "transportation", no matter whether it is an individual, corporation, government office, etc.
IEA's data and those compiled by the UN Statistics Bureau are for-fee publications, though someone with access could extract the relevant numbers for us here.