LCFS - Screen Readers Only Announce Bold Text Instead of Complete Text Content

**Describe the Bug**
Screen readers are only reliably announcing text that is formatted in bold. Non-bold text is not consistently read unless the user manually navigates to and selects individual words. This issue appears across multiple areas of the application and prevents screen reader users from accessing content in a logical and efficient manner.

**Expected Behaviour**
Screen readers should announce all text content within page elements, including headings, labels, instructions, paragraphs, tables, and form content, regardless of whether the text is bold or non-bold. Users should be able to navigate and have content read naturally without selecting individual words.

**Actual Behaviour**
Screen readers only announce bold text consistently. Non-bold text is often skipped or requires users to select individual words before it can be read, resulting in incomplete access to content.

**Implications**
This creates a significant accessibility barrier for users who rely on screen readers and may prevent them from understanding page content, instructions, and system information. The issue negatively impacts usability, navigation, and compliance with WCAG 2.2 accessibility requirements.

**Steps To Reproduce**
Steps to reproduce the behaviour:

User/Role: Screen Reader User

1. Go to any page within the application containing a mix of bold and non-bold text.
2. Launch a supported screen reader (e.g., NVDA, JAWS, or VoiceOver).
3. Navigate through the page using standard screen reader reading commands.
4. Observe that bold text is announced, while non-bold text is not consistently read.
5. Attempt to access the non-bold text.
6. Note that individual words may need to be selected before the content is announced by the screen reader.

Start by reproducing the issue on pages containing mixed bold and non-bold text with a supported screen reader such as NVDA, JAWS, or VoiceOver. Compare reading behavior for headings, labels, instructions, paragraphs, tables, and forms. Done when all page text is announced naturally without selecting individual words; no file or test is named.

Written by the indexing model from the issue text.