Where does investment originate from in the circular flow model?

Your 2 questions are very relevant because are part of the inconsistencies included in conventional economics, which are especially visible in introductory texts.

1. What does it mean "Investment" if it flows from financial markets to firms, through goods markets? In fact, in the context you found it, Investment is one of the sources of the mass of aggregate income (all the flows in your picture are monetary flows): the payment for the capital-goods purchased (therefore, counterbalance of one of the destinies for the goods produced). That's the meaning of "I" in the famous equation Y=C+I (+G+NX). In fact, your picture is very interesting in trying to incorporate the details of the "transformation" of savings "S" into investments "I". In fact, S is a destination for income, whilst I is a form of expense. But this is done in a manner that introduce more inaccuracies than enlightenments, because, as goods, investments go from producers/sellers (firms) to investors/buyers (other firms); as money, the payment of those goods has, of course, the same actors with opposite direction (this is the "I" in your scheme), and there still are as financial investments as in "financial markets", whose final net flow (monetary) is from capital holders (households as well as firms) to capital borrowers (firms, to keep the matter at basics). In summary: that picture is a very good one regarding the relations among different monetary flows (the arrows), but is not accurate at all regarding the participating agents (the boxes). Also consider to have a look on the alternative scheme that @Giskard pointed in his comment, which is also interesting. It's not a quasi-dynamic cycle, but a clasification of monetary flows and their non-monetary counter-flows (except for "financial markets", where both flows are monetary) with also 3 markets, but only 2 agents. Sadly, it doesn't include any "real investment" ("I") so it also doesn't shed light on your question.

2. How does wealth ever actually increase/the economy ever grow if everything is a cycle? As yourself stated, On a more basic level thus even a more relevant question. The basic answer is: <<if things worked as in that scheme, in fact there could be no changes, except from "rest of the world": this is the only "exogenous" source stream>>. The complete answer must include the fact that conventional macroeconomics rests on a static point of view, considering economies as static (as opposite to dynamic), and taking temporal change only as "jumps" between different static (a-temporal) states. For example: increase in GDP in 1990 respect to 1989 is the jump of a situation lasting from 1-Jan to 31-Dec in 1989 to a new situation along all the following year.

You will find that those kind of absolutely relevant questions are generally and plainly not answered, deprecated as non-relevant. When they are, exceptionally, taken into consideration, the answers are of an elusive nature, because conventional macroeconomics have no explanation. That's why you find answers like << the income increases each year as economy becomes more productive thanks to improvements in technology >>, which, the more you think about it, the less is explaining: "production", "technology" or "value" are dimensions not present in the scheme, and before we could take them into the explanation we have to carefully explain them and explain how they exactly are present in the scheme.