As noted earlier, Richard wonders "why we insist on having the word 'novel' encompass so much. Why must it be asserted that the books written by Sebald and Bolaño 'are certainly' novels? Are they?"
Well, the actual is real and all that, so, yes, Sebald and Bolaño's books are novels because that is what we call "a fictitious prose narrative of book length"! I don't think we should worry too much about how restrictive the term might be because, in practice, it has always been a wonderfully capacious and imprecise thing. And a perenially contested term to boot: indeed, each and every new work of art potentially contests it...
According to my Shorter OED, short stories of the type contained in works like the Decameron and Heptameron were being called novels by 1566. By 1643 the definition had morphed to be "a fictitious prose narrative of considerable length, in which characters and actions representative of real life are portrayed in a plot of more or less complexity."
What is going on in writing in English over this time? Thomas Malory's Le Morte d'Arthur, (written circa 1470) was published in 1485, Philip Sidney's The Countess of Pembroke's Arcadia in 1581, the first English translation of Don Quixote was 1620, Bunyan's The Pilgrim's Progress 1678, Defoe's Robinson Crusoe 1719, Swift's Gulliver's Travels 1726 and Samuel Richardson's Pamela 1740 (dates thanks to Wikipedia). Sterne's Tristram Shandy first appeared in 1759 two years after the OED says the word novel was being used to describe "this type of literature."
The problem with the term, as Richard points out, following Josipovici, is that the Victorian novel became fused and confused with all that fiction could be ("Josipovici has argued that the narrative mode of the 19th century novel became so dominant... that we expect it to hold true for very different sorts of narratives") but, as we've seen, the coinage pre-dates the Victorian era by hundreds of years (Victoria reigned 1837-1901).
In The True Story of the Novel, Margaret Doody argues -- to quote the blurb -- against the "conventional view of the novel, arguing that instead of being the defining achievement of the English middle class, the novel is an older more cosmopolitan creation, a protean form that emerged from the ancient cultures of Africa, Asia and Europe." Doody says, "One of the most successful literary lies is the English claim to have invented the novel.... One of the best-kept literary secrets is the existence of novels in antiquity." Doody is arguing against Ian Watt whose influential Rise of the Novel: Studies in Defoe, Richardson and Fielding (1957) traced the rise of the modern novel "to philosophical, economic and social trends and conditions that become prominent in the early 18th century". But we can recognise the truth in Watt's history of the modern British novel whilst accepting Doody's corrections to his parochialism: "novels" pre-date novels, considerably pre-date Victorian novels, certainly pre-date the use of the word novel in England, and have, for sure, existed in many forms in many cultures (the work of Franco Moretti, the novel as a "planetary form" and all that, is important here).
What I don't know is "why a term of art derived from the French word for 'new' under a very historically contingent set of circumstances" actually arose -- what was deemed to be so new? Presumably -- and Josipovici has argued something along these lines -- Epic poetry, mystery/miracle plays and folk tales no longer functioned as successfully as they once did and what was new was that storytelling was becoming writing. Is that right?