That is, what OS was the first to abstract the external storage away from physical volumes at the OS level, even by pre-allocated fixed-size chunks, by introducing the notion of a "personal virtual volume" or similar, according to metadata maintained by the OS?
My second guess is CTSS. It was operational in 1961, but at that time had only tapes for user file storage. I suppose that tape name records don't constitute 'metadata' in the sense required by this question.
A disk was added somewhere around 1962 to 1963. The CTSS Programmer's Guide from 1963 mentions
- the installation of the IBM 1301 disk file; and 5) the design and programming of a master disk control subroutine (memo CC-196) and an associated disk editor program (memo CC-208)
I have not found CC-196 online, but CC-208 describes the Master File Directory (MFD), User File Directories (UFDs). Each UFD entry contains the starting track number and number of tracks for the file; I infer files are contiguous.
(Page 1 if you want to look it up; it doesn't cut and paste well)
So, this definitely qualifies as a file system within the requirements of the question. It also demonstrates a pleasing ancestral relationship to the subsequent hierarchical filesystem paper. Whether CTSS was 'first' depends on whether anyone else here can find anything earlier.
(This answer has now been determined not to satisfy the now-clarified requirements of the question. Nevertheless, the discussion seems useful, if only in my own mind, so I will leave it here. But see my other answer about CTSS.)
I will guess that the standard answer for 'first' applies here: the Atlas Supervisor.
Section 6 of the linked document talks about data handling.
The fast computing speed of Atlas and the use of multiple input and output peripheral equipments enable the computer to handle a large quantity and variety of problems. These will range from small jobs for which there is no data outside the program itself, to large jobs requiring several batches of data, possibly arriving on different media. Other input items may consist of amendments to programs, or requests to execute programs already supplied. Several such items may be submitted together on one deck of cards or length of punched tape. All must be properly identified for the computer.
To systematise this identification task, the concept of a document has been introduced. A document is a self-contained section of input information, presented to the computer consecutively through one input channel. Each document carries suitable identifying information (see below) and supervisor keeps in the main store a list of the documents as they are accepted into the store by the input routines, and a list of jobs for which further documents are awaited.
This is perhaps more dynamic than you had in mind; the supervisor only maintains identification/location information for 'active' files. However, I think this is not so very different to systems using exchangeable disc storage; the OS often only knows the content of what's currently mounted online - unless of course it has the design that maintains a single catalogue for all volumes.
The important feature that makes this a valid answer is that the user assigns a name to the document, the program asks for the document by name, and the supervisor uses the name to match the program request to the hardware on which the document lives (which might, transparently, be on magtape if input spooling is being used).