Thanks for the prompt response.
I'm going to largely ignore the part about gathering evidence for its use for now:
Then there's the question of what the use case is for these characters outside the legacy environment. You must have played with these for some purpose, but it seems that you are not using them as part of a notational system or part of an established user community with its own conventions. That means, so far, we have to assume these aren't really used, except by a few people to toy around.
Gathering evidence for use is difficult. For one, search engines aren't geared towards searching for sets of Unicode characters (or even single Unicode characters).
The ZX Spectrum does appear to be having a revival at the moment, but I have not attempted, e.g. to trawl message boards for anyone discussing block elements in Unicode.
However, the larger problem is that if the code block is—as I am putting the case forward for here—incomplete, then this will have severely curtailed its use, and have prevented the establishment of communities using the characters.
For example in plain text messaging (e.g. SMS, Twitter, Facebook, and many email clients), the lack of a "blank" makes the use of block elements awkward at best, limiting any community which might form organically.
The systems that used these block elements had fixed display cells, and a space character had the width of a display cell.
Yes, but now it's intermixed with other text, as it's in Unicode.
...the width could, in principle, be determined by referencing the surrounding characters, instead of having to be given as part of the input (i.e. by using a special character code).
No, that's impossible, as spaces at the start, end, or middle may refer to a normal word space, or may refer to a block element blank. Trying to tell from context would be guesswork at best.
For example, to go back to the simple example of ▝▞ ▞▞. It's ambiguous, and may be interpreted in different ways:
1. if interpreted as a normal space, it may refer to two chunks of block characters: "▝▞" and "▞▞", with a space separator.
2. if the space is interpreted as a block-element blank, it would be seen as a single graphical unit, which can only be rendered correctly using markup:
I use the term "blank" rather than "space", as this usage is not semantically a space and should not be a breaking element. The Braille block similarly has a Blank (i.e. "no dots raised"), which may be rendered with no glyph, but is not semantically a space, even though it may come from systems which originally encoded it as one. The Block Elements block is incomplete without a blank.
Adding new space characters would be the last resort, because it would invalidate so much software: you may well want software to continue to identify U+0020 for purposes like word breaking (and use U+00A0 when that is undesirable), for example.
It wouldn't invalidate any software any more than adding any other Unicode element, other than highly specialized software which would otherwise have no way of differentiating the intention of the author: whether they meant for a space to be a space or for it to be a blank in a semigraphical construct.