opensnoop -F resolves full paths outside container mount namespace when using relative paths

**Description:**

When using `bcc`'s `opensnoop` with the `-F` (full path) option in a containerized environment, I observed an unexpected behavior regarding relative file paths accessed within the container. Specifically, when a process inside a container opens a file using a *relative path* in a *mounted volume*, `opensnoop -F` resolves the full path **using the host's mount namespace**, not the container's.

**Steps to Reproduce:**

1. Start a container with a volume mounted, e.g., `-v /host/data:/container/data`.
2. Inside the container, run a process that accesses a file using a relative path, such as:
```bash
cd /container/data
cat file.txt
```
3. On the host, run:
```bash
sudo opensnoop -F
```

**Expected Behavior:**

The full path should resolve based on the container's mount namespace — for example:
```
/container/data/file.txt
```

**Actual Behavior:**

`opensnoop -F` reports the full path as seen from the *host* perspective — for example:
```
/host/data/file.txt
```

This can be misleading when tracing file activity scoped to containers, as it breaks the container's isolation abstraction and complicates debugging or monitoring within container contexts.