Linter doesn't flag usage of manual identity object instead of `I` glyph

The language has a glyph for the identity object — I, which is
[x] > (φ ↦ x). Still, we keep writing that object out by hand. In
objectionary/eo both printf.eo and as-decimal.eo carried a hand-made
one-attribute formation that a glyph could spell, and nothing complained
(objectionary/eo#7822). I'd like a lint that catches an identity object
wherever it exists and suggests replacing it with I.

Incorrect:

# Jeff.
[] > jeff
  x > [x] >> id
  foo id > @

# Jeff.
[] > jeff
  [value] > pass
    value > @
  foo pass > @

Correct:

# Jeff.
[] > jeff
  foo I > @

What to catch

A formation is the identity object when it binds exactly one void, decorates
that very void, and holds nothing else. In XMIR that is:

<o name="id">
  <o name="x" base="∅"/>
  <o name="φ" base="ξ.x"/>
</o>

The predicate already exists, in eo:identity of eo's
eo-printer/src/main/resources/org/eolang/printer/print/to-eo-tree.xsl
(objectionary/eo#6834), where the printer uses it to print any such formation
back as the glyph:

eo:abstract($o) and not(eo:has-data($o)) and count($o/o) = 2
  and eo:void($o/o[1]) and empty($o/o[1]/(@local, @type, @args))
  and $o/o[2]/@name = $eo:phi and empty($o/o[2]/o)
  and empty($o/o[2]/@const)
  and $o/o[2]/@base = concat($eo:xi, '.', $o/o[1]/@name)

The tight tests there are load-bearing and worth keeping: exactly two children,
because a third is an argument applied to the formation and the glyph has
nowhere to put it; and the φ base rooted at ξ, because an unrooted name
denotes some other object entirely.

The one catch

I desugars to precisely the same XMIR — eo's Emissions.identity emits a
baseless object, a void named x, and a φ bound to it — so a lint that matches
the shape alone would flag every use of the glyph, I > sanitized included.
Two ways out, and I have no strong preference:

1. Have eo's parser mark the desugared glyph (an attribute on the <o>), and
   let the lint report the shape only where the mark is absent. Cleanest, but it
   needs a change in eo first.
2. Tell them apart by position: for the desugared glyph the formation, its void
   and its φ all carry the same @line and @pos, since all three come from
   the one character; a hand-written formation spreads over more than one line
   or more than one column. No change in eo needed, at the price of a heuristic.

Deliverables

• src/main/resources/org/eolang/lints/misc/identity-object.xsl — warning
  severity, and maybe experimental="true" at first, the way
  design/no-attribute-formation.xsl does it
• src/main/resources/org/eolang/motives/misc/identity-object.md
• a test pack under
  src/test/resources/org/eolang/lints/packs/single/identity-object/, with a
  catching case and a no-catch case — the no-catch one guarding against the
  false positive above, plus the near misses the predicate must let through: a
  formation with two voids, one whose φ is applied to something, and one whose φ
  reaches a name that isn't its void