Add comment explaining the need for separating postinstall from install

There are two reasons why we need to run `yarn install --frozen-lockfile` with
the `--ignore-scripts` flag and run `yarn run postinstall` afterwards:
1. Sometimes during installation of the `monitoring` project we were getting
   random failures related to the presence of the `prepare-dependencies.sh`
   script from `threshold-network/solidity-contracts` in the project
   subdependencies. Lately we managed to remove the script from the
   `development`-tagged packages, but the script is still present in `goerli`
   and `mainnet` packages. As `monitoring` project installs `goerli` and
   `mainnet` `@keep-network/tbtc-v2` packages during `yarn install`, the random
   failures may still occur. As a workaround for this problem we will ca run
   install step with the `--ignore-scripts` flag which installs the dependencies
   without running the postinstall scripts (neither in subdependencies nor in
   the main project). We then add a step that executes `yarn run postinstall` -
   this way we still execute any postinstall scripts from the main project
   (currently it's `npm rebuild bcrypto`) script.
2. The `bcoin` holds their `node_modules` in VCS
   (https://github.com/bcoin-org/bcoin/tree/master/node_modules). That means
   `bcrypto` dependency is already here when we run `yarn install`. For an
   unknown reason, `yarn install` does not build all native components in that
   case, hence the need for manual `postinstall` that does it again under the
   hood.

monitoring/Dockerfile

@@ -24,6 +24,8 @@ COPY package.json yarn.lock ./
 COPY tsconfig.json ./
 COPY src ./src
 
+# Run install and postinstall separatelly to avoid problems with `bcrypto` and
+# `install-dependencies.sh`.
 RUN yarn install --frozen-lockfile --ignore-scripts
 RUN yarn run postinstall