As Arturo's comment notes, this matrix is the transpose of the companion matrix associated with a certain polynomial. In particular, we have $M = C(p)^\top$ with
$$
p(x) = x^m - 2x^{m-1} + 2x^{m-2} - \cdots -2x^2 + 2x - 1.
$$
With that, we find that $M^{2m} = I$ will hold if and only if $p(x)$ divides $x^{2m} - 1$. We find that this is indeed the case. In particular, we find that $x^{2m - 1} = p(x)q(x)$, where
$$
q(x) = -p(-x) = x^{m} + 2x^{m-1} + \cdots + 2x + 1.
$$
We can see this as follows.
\begin{align}
q(x)p(x) &= (x^{m} + 2x^{m-1} + \cdots + 2x + 1) (x^m - 2x^{m-1} + 2x^{m-2} - \cdots -2x^2 + 2x - 1)
\\ & = [(x^m + x^{m-1}) + (x^{m-1}+x^{m-2}) + \cdots + (x^1 + x^0)]
\\ &\qquad \cdot [(x^m - x^{m-1}) - (x^{m-1} - x^{m-2}) + \cdots +(x^1 - x^0)]
\\ & = (x+1)[x^{m-1} + \cdots + x + 1] \ \cdot\ (x-1)[x^{m-1} - x^{m} + \cdots - x + 1]
\\ & =
(x-1)[x^{m-1} + \cdots + x + 1] \ \cdot\ (x+1)[x^{m-1} - x^{m} + \cdots - x + 1]
\\ & = (x^m - 1)(x^m + 1) = x^{2m} - 1.
\end{align}
Alternatively, we can use complex numbers. Let $\omega = e^{\pi i/m}$, i.e. an elementary $2m$-th root of unity. We find that
$$
q(x) = (x + 1) \prod_{k=1}^{m-1}(x - \omega^{2m}),
$$
so that its roots are $-1$ along with $\omega^{2k}$ for $k = 1,\dots,m-1$. On the other hand, the roots of $p(x) = -q(-x)$ are $-1$ together with $-\omega^{2k}$ for $k = 1,\dots,m-1$. Because $-\omega^{2k} = \omega^{2k + m} = \omega^{2k - m}$, we can conclude that the roots of $p(x)q(x)$ are equal to all of the $2m$-th roots of unity, each with multiplicity $1$.
Or, we could use these facts to see that the roots of $p$ are indeed roots of $x^{2m} - 1$, which implies that $p$ divides $x^{2m} - 1$.
Another proof can be given by identifying the eigenspaces of $M^m$, where $M$ is considered acting on the column vectors with $m$ coordinates by multiplying from the left.
Let $v_1, v_2, \cdots, v_m$ be the column $1,2,\cdots, m$ of the following matrix respectively, where hidden entries are zeros.
$$\pmatrix{
& & & & &1&1\\
& & & &1&1& \\
& & &\vdots&1& & \\
& &1&\vdots& & & \\
&1&1& & & & \\
1&1& & & & & \\
1& & & & & &-1\\
}$$
Verify that $Mv_1=v_2$, $Mv_2=v_3$, $\cdots$, $Mv_{m-2}=v_{m-1}$ as well as $Mv_{m-1}=v_m$ and $Mv_m=-v_1$.
So $M^mv_1=-v_1$.
Hence $M^mv_i=-v_i$ for all $i$.
Hence $M^{2m}v_i=v_i$ for all $i$.
Let $u=\pmatrix{1\\1\\\vdots\\1}$. Verify that $Mu=u$.
Hence $M^{2m}u=u$.
Since $\pmatrix{v_1, v_2,, \cdots, v_{m-1}, u-v_1-v_3-\cdots-v_{m-2}}$ is the following triangular matrix, which is nonsingular,
$$\pmatrix{
& & & & &1&1\\
& & & &1&1& \\
& & &\vdots&1& & \\
& &1&\vdots& & & \\
&1&1& & & & \\
1&1& & & & & \\
1& & & & & &\\
}$$
$v_1, v_2, \cdots, v_{m-1}, u$ are linearly independent vectors. Since there are $m$ of them, $M^{2m}$ fixes the entire space of column vectors with $m$ coordinates. Hence $M^{2m}=I$.