TritonContext.simplify() does not eliminate dead code completely

[albertlab0]

Consider the 32 bit code code.zip

Using the following code for dead code elimination:
sblock = ctx.simplify(block) ctx.disassembly(sblock, addr)

(rebase the code to 0x27b20)

before code elimination:

    0x27b20:	push	ebp	(1 bytes: 55)
    0x27b21:	mov	ebp, esp	(2 bytes: 8bec)
    0x27b23:	and	esp, 0xfffffff0	(3 bytes: 83e4f0)
    0x27b26:	push	esi	(1 bytes: 56)
    0x27b27:	push	edi	(1 bytes: 57)
    0x27b28:	push	ebx	(1 bytes: 53)
    0x27b29:	sub	esp, 0x144	(6 bytes: 81ec44010000)
    0x27b2f:	rol	ecx, cl	(2 bytes: d3c1)
    0x27b31:	xchg	ecx, edx	(2 bytes: 87d1)
    0x27b33:	neg	dl	(2 bytes: f6da)
    0x27b35:	inc	cl	(2 bytes: fec1)
    0x27b37:	shl	eax, cl	(2 bytes: d3e0)
    0x27b39:	dec	dh	(2 bytes: fece)
    0x27b3b:	xchg	ecx, edx	(2 bytes: 87d1)
    0x27b3d:	xchg	edx, eax	(1 bytes: 92)
    0x27b3e:	shr	ch, 1	(2 bytes: d0ed)
    0x27b40:	mov	al, cl	(2 bytes: 8ac1)
    0x27b42:	xchg	edx, ecx	(2 bytes: 87ca)
    0x27b44:	movzx	edx, dh	(3 bytes: 0fb6d6)
    0x27b47:	inc	eax	(1 bytes: 40)
    0x27b48:	not	eax	(2 bytes: f7d0)
    0x27b4a:	adc	dh, dl	(2 bytes: 12f2)
    0x27b4c:	inc	dh	(2 bytes: fec6)
    0x27b4e:	xor	ch, dh	(2 bytes: 32ee)
    0x27b50:	inc	cl	(2 bytes: fec1)
    0x27b52:	inc	edx	(1 bytes: 42)
    0x27b53:	rcr	dl, 6	(3 bytes: c0da06)
    0x27b56:	sub	edx, edx	(2 bytes: 2bd2)
    0x27b58:	shl	ch, 3	(3 bytes: c0e503)
    0x27b5b:	lea	ecx, [edx + edx]	(3 bytes: 8d0c12)
    0x27b5e:	adc	cl, byte ptr [esp + 0xc]	(4 bytes: 124c240c)
    0x27b62:	adc	cl, ch	(2 bytes: 12cd)
    0x27b64:	sbb	dl, ch	(2 bytes: 1ad5)
    0x27b66:	rcr	dh, 7	(3 bytes: c0de07)
    0x27b69:	movzx	ecx, cl	(3 bytes: 0fb6c9)
    0x27b6c:	inc	ecx	(1 bytes: 41)
    0x27b6d:	inc	edx	(1 bytes: 42)
    0x27b6e:	movzx	eax, dh	(3 bytes: 0fb6c6)
    0x27b71:	inc	ch	(2 bytes: fec5)
    0x27b73:	sbb	eax, eax	(2 bytes: 1bc0)
...

After code elimination：

0x27b20: push ebp
0x27b21: mov ebp, esp
0x27b23: and esp, 0xfffffff0
0x27b26: push esi
0x27b27: push edi
0x27b28: push ebx
0x27b29: sub esp, 0x144
0x27b2f: rol ecx, cl
0x27b31: xchg ecx, edx
0x27b33: neg dl
0x27b35: inc cl
0x27b37: shl eax, cl
0x27b39: dec dh
0x27b3b: xchg ecx, edx
0x27b3d: xchg edx, eax
0x27b3e: shr ch, 1
0x27b40: xchg edx, ecx
0x27b42: movzx edx, dh
0x27b45: adc dh, dl
0x27b47: inc dh
0x27b49: xor ch, dh
0x27b4b: inc cl
0x27b4d: inc edx
0x27b4e: rcr dl, 6
0x27b51: sub edx, edx
0x27b53: shl ch, 3
0x27b56: lea ecx, [edx + edx]
0x27b59: adc cl, byte ptr [esp + 0xc]
0x27b5d: adc cl, ch
0x27b5f: sbb dl, ch
0x27b61: rcr dh, 7
0x27b64: movzx ecx, cl
0x27b67: inc ecx
0x27b68: inc edx
0x27b69: movzx eax, dh
0x27b6c: inc ch
0x27b6e: sbb eax, eax
0x27b70: movzx eax, dl

Here, noticed that
0x27b51: sub edx, edx

so all the dl, edx related code before this should be eliminated as dead code.
And furthermore, since 0x27b56: lea ecx, [edx + edx]
ecx is 0, ecx, and cl related code before 0x27b56 should be eliminated as dead code as well