eval(Meta.parse(repr(x))) != x for some Symbols

[mhauru]

As an offshoot of #60428, I noticed that there are a lot of one-character Symbols whose repr does not parse back as the same Symbol:

julia> for i in 0:typemax(UInt32)
           c = nothing
           try
               c = Char(i)
               sym = Symbol(c)
           catch e
               continue
           end
           pass = (eval(Meta.parse(repr(sym))) == sym)
           if !pass
               println("fail: Char($i) ($(repr(c)))")
           end
       end
fail: Char(181) ('µ')
fail: Char(183) ('·')
fail: Char(603) ('ɛ')
fail: Char(884) ('ʹ')
fail: Char(903) ('·')
fail: Char(2392) ('क़')
fail: Char(2393) ('ख़')
fail: Char(2394) ('ग़')
fail: Char(2395) ('ज़')
fail: Char(2396) ('ड़')
fail: Char(2397) ('ढ़')
fail: Char(2398) ('फ़')
fail: Char(2399) ('य़')
fail: Char(2524) ('ড়')
fail: Char(2525) ('ঢ়')
fail: Char(2527) ('য়')
fail: Char(2611) ('ਲ਼')
fail: Char(2614) ('ਸ਼')
fail: Char(2649) ('ਖ਼')
fail: Char(2650) ('ਗ਼')
fail: Char(2651) ('ਜ਼')
fail: Char(2654) ('ਫ਼')
fail: Char(2908) ('ଡ଼')
a
n
d

m
a
n
y

m
a
n
y

m
o
r
e

This seems quite contrary to spirit of what show should do.

[ararslan]

A few of those are intentional:

julia/base/strings/unicode.jl

Lines 177 to 192 in 1a6ed25

	"""
	`Dict` of `original codepoint => replacement codepoint` normalizations
	to perform on Julia identifiers, to canonicalize characters that
	are both easily confused and easily inputted by accident.
	!!! warning
	When this table is updated, also update the corresponding table in `src/flisp/julia_charmap.h`.
	"""
	const _julia_charmap = Dict{UInt32,UInt32}(
	0x025B => 0x03B5, # latin small letter open e -> greek small letter epsilon
	0x00B5 => 0x03BC, # micro sign -> greek small letter mu
	0x00B7 => 0x22C5, # middot char -> dot operator (#25098)
	0x0387 => 0x22C5, # Greek interpunct -> dot operator (#25098)
	0x2212 => 0x002D, # minus -> hyphen-minus (#26193)
	0x210F => 0x0127, # hbar -> small letter h with stroke (#48870)
	)

Unsure about the rest.

[mlechu]

I think all of these are being normalized (Base.JuliaSyntax.normalize_identifier(string(c)) != string(c)), but it's still a printing bug if str1 normalizes to a different str2 and we print Symbol(str1) with :-syntax instead of Symbol("")-syntax