I'm a UX designer on the Metabase core team, and I'll try to answer some of the bullet points about Metabase as honestly as I can from my perspective. Also happy to answer any other specific questions folks might have.
One of the main things we're pretty proud of about Metabase is the care we've tried to take with its design and usability. We often get positive feedback from users about the ease of using Metabase and its aesthetic appeal. Beyond that, we have a simple GUI querying interface that's great for users who don't know how to write SQL or native queries. Lastly, we've been doing some interesting things using ML to provide automatic insights called x-rays to show things like growth analysis for time series at the press of a button.
In terms of scaling, we have a wide range of companies using Metabase with either very small hobby DBs or huge Redshift instances, both with success. We have some work to do in making sure super massive DBs don't overly slow down certain parts of the app (like field mapping with SQL filters, or paginating long lists more efficiently). In terms of scaling in-app files or organizing things, Metabase lets you put your saved queries into collections, which you can set user group permissions on.
I think Metabase checks off most of the common visualization options: line, area, bar, row chart, pie/donut, scatter/bubble plot, region map, pin map, progress bar, and funnels. Two requests that we get often that we're working on are line+bar combo charts and cohort analysis tables (which can be accomplished with a pivot table, but without nice color formatting).
Under the hood, we've been trying to add a lot of power over the last year: data access controls, field value remapping, nested or sub-queries, SQL/native query variables, the ability to embed charts or dashboards in other applications, fine-grained scheduling controls for DB sync and analysis, and Google auth and LDAP login.
Not sure if you have any specific questions about DB or schema requirements, but I'll just say we currently support: Postgres, MySQL, Druid, SQL Server, Redshift, MongoDB, Google BigQuery, SQLite, H2, CrateDB, Oracle, Vertica, and Presto. There's also an experimental Google Analytics driver that's not 100% fully featured, but is still quite good (we use it to visualize a few of our own metrics).
Lastly, our team is quite active on our GH repo (https://github.com/metabase/metabase/), doing our best to quickly respond to issues, requests, and PRs.
I hope that helps.