Use ArrayVec::push_many_from_slice to make code safer

We were ignoring potential (even though it shouldn't happen) failure.
The new way however is correct by construction.

The `ArrayVec` method is from bluss/arrayvec#237

Author: Ericson2314

Cargo.toml

@@ -13,6 +13,9 @@ nanos_sdk = { git = "https://github.com/LedgerHQ/ledger-nanos-sdk.git" }
 nanos_ui = { git = "https://github.com/LedgerHQ/ledger-nanos-ui.git" }
 ledger-parser-combinators = { git = "https://github.com/alamgu/ledger-parser-combinators", branch="async-split-take-2" }
 
+[patch."crates-io".arrayvec]
+git = "https://github.com/obsidiansystems/arrayvec"
+branch = "push-many-from-slice"
 [patch."https://github.com/LedgerHQ/ledger-nanos-sdk.git".nanos_sdk]
 git = "https://github.com/alamgu/ledger-nanos-sdk.git"
 branch = "relocating-loader-w-fixes"

src/lib.rs

@@ -363,8 +363,8 @@ impl Readable for ByteStream {
                     Err(_) => reject().await,
                 };
                 let avail = &chunk[self.current_offset + HASH_LEN..];
-                let consuming = core::cmp::min(avail.len(), buffer.remaining_capacity());
-                buffer.try_extend_from_slice(&avail[0..consuming]).ok();
+                let remaining = buffer.push_many_from_slice(&avail).len();
+                let consuming = avail.len() - remaining;
                 self.current_offset += consuming;
                 if self.current_offset + HASH_LEN == chunk.len() {
                     self.current_chunk = chunk[0..HASH_LEN].try_into().unwrap();